NAYACAKALOU MEDAL ADDRESS
EXPLAINING THE ABERRANT AUSTRONESIAN LANGUAGES
OF SOUTHEAST MELANESIA: 150 YEARS OF DEBATE
ANDREW PAWLEY
Australian National University
This paper reviews the history of ideas about the origins and development
of the Melanesian languages, particularly those languages of Southeast
Melanesia that have been called “aberrant”. Following tradition I use the
phrase “Melanesian languages” to refer to languages of Melanesia that belong
to the Austronesian family but are not Polynesian. The Melanesian languages,
which number more than 450 (see note 4), do not constitute a separate branch
of Austronesian apart from the Polynesian and Micronesian groups. The
usefulness of the term is chiefly that it defines a large set of languages that
are present in a particular geographic region, many of which have certain
attributes that have long puzzled historical linguists.1
It is a striking fact that a high proportion of Melanesian languages
are atypical in certain respects when compared with the better-known
Austronesian languages of Melanesia, Polynesia and the western Austronesian
region. The label “aberrant” has been applied to such languages by George
Grace, who contrasts them with “exemplary” or “well-behaved” Melanesian
languages (Grace 1981, 1992, 1996).2 He emphasises that we are dealing here
with a continuum rather than with two sharply distinct types.
Grace observes that a language or group of languages (call it L) can be
aberrant in any of the following respects.
(i) L shows relatively few cognates with other Austronesian languages.
(ii) L’s sound system departs radically from the systems reconstructed for
early stages of Austronesian, often making cognates hard to identify.
(iii) L’s grammatical structure is atypical.
(iv) It is unusually difficult to apply the Comparative Method to L because
multiple sound correspondences obscure the distinction between directly
inherited forms and borrowed forms.
These four variables are not necessarily linked but some languages exhibit
two or more of characteristics (i)-(iv).
215
216
Explaining the Aberrant Austronesian Languages
Table 1 provides an example of the extent to which Melanesian languages
can vary in the number of cognates they share in basic vocabulary. The
variation can be conveniently measured in terms of how many words are
retained from the Austronesian inter-stage known as Proto Oceanic (POc),
ancestral to most of the Austronesian languages of the Pacific Islands
including all the Melanesian languages except about 30 spoken at the western
end of New Guinea (see 1.4). The first line of the table shows POc words for
six concepts that are part of core basic vocabulary, words that languages tend
to retain for several millennia. Each POc word is a well-attested reconstruction
and most of these etyma are continued (though often with sound changes)
in more than half of the contemporary Oceanic languages. Below the POc
word we find the most common word for that concept in eight contemporary
Oceanic languages. Five of these languages are spoken in the Solomon
Islands: Ririo on Choiseul, Zabana and Maringe on Santa Isabel, Gela on
Florida and Arosi on Makira.3 The other three are Standard Fijian, Tongan
and Samoan. Words in daughter languages that are not cognate with the POc
etymon are shown in italics.
The three lexically innovative languages in the sample are Ririo, Zabana and
Maringe. Ririo keeps just one, and Zabana and Maringe two of the POc etyma.
By contrast, Standard Fijian, Tongan and Samoan each retain all six etyma,
Gela five and Arosi four. Of course, a sample of six words is tiny but a larger
sample would yield quite similar results for these particular languages.
Table 1: A comparison of reflexes of six Proto Oceanic words in six languages
of Melanesia and in Tongan and Samoan.
English
arm
ear
liver
bone
skin
louse
POc
*lima
*taliŋa
*qate
*suRi
*kulit
*kutu
Ririo
Zabana
Maringe
Gela
karisi
ŋgel
lima
khuli
khebu
kariŋa
rogo
kame
lima
taliŋa
kuli
Arosi
rima
Fijian
liŋa
daliŋa
Tongan
lima
Samoan
lima
tutuen
punda
kapat
knubra
guli
kola
huma
ate
huli
kafu
gui-guli
utu
gutu
theli
gutu
kote
su-suri
ʔuri-ʔuri
ate
sui
kuli
teliŋa
ʔate
hui
kili
kutu
taliŋa
ate
ivi
ʔili
ʔutu
kutu
Andrew Pawley
217
It can also be seen in Table 1 that, although there are some mutations in
form, most of the cognate forms in the daughter languages have not departed
far from the POc original and consequently cognates are easily recognised,
even in Ririo, Zabana and Maringe. This suggests that none of the eight
daughter languages in this sample is markedly aberrant in its sound system,
although the data in this table are too scanty to speak with confidence.
Table 2 shows a different situation: reflexes of five POc words in four
languages of northern New Caledonia that have undergone quite far-reaching
sound changes. The New Caledonian forms are all good cognates but some
are not easily recognisable as such at first glance. The table is adapted from
Ozanne-Rivierre (1992:193).
Table 2: Reflexes of Proto Oceanic forms in four New Caledonian languages.
English
daytime
hear
blood
cooking pot
stand up
POc
*raqani
*roŋoR
*draRaq
*kuron
*tuqur
Yuanga
ṭɛɛ
ṭone
(ku)ṭa
Kumak
taan
tãlã
ndaa-
cet, cela-
kuut
Pwapwâ
(mavi)taan
tena
(ku)taa-
cit-, cile-
cuut
Cèmuhî
tàn
tēne
ilà
cūut
kɔɔ
When a language or language group L differs radically in lexicon and/or
phonology and/or grammar from its relatives M and N, which are quite
similar to one another, four broad types of historical explanation are normally
considered.
(a) Genealogical diversity. M and N are much more closely related to each
other than to L, meaning that L has undergone a relatively long period of
separate development since it diverged from M and N.
(b) Contact-induced change. L has been extensively changed by borrowing from
a language or languages that were previously of a very different cast to L.
(c) An exceptional degree of organic (system internal) linguistic change.
Particular linguistic structures are subject to natural processes in speech
production and comprehension, which allow children to reinterpret these
structures. L has undergone more far-reaching innovations of this type
than M and N.
218
Explaining the Aberrant Austronesian Languages
(d) An exceptional degree of socially-induced linguistic change. L has changed
more rapidly than M and N because of the far-reaching effects of social
processes internal to the community, e.g., name taboo, the desire to be
different from its neighbours, the emergence of a prestige dialect, or the
lack of clear linguistic norms constraining language change.
The occurrence of population bottlenecks owing to migration, warfare,
disease, etc. perhaps falls into category (d) insofar as it is associated with
social disruption and the loss of linguistic knowledge. Otherwise it is simply
a demographic circumstance that would allow innovations, whatever their
source, to spread quickly through the speech community.
In the case of the aberrant Melanesian languages, each of these four main
types of explanations has been proposed. In addition, a fifth, less orthodox
type was for a long time favoured by some scholars, namely,
(e) Pidginisation. This supposes that L is not a true sister language to M and
N. Instead L was created as a pidgin, a hybrid formed when speakers of
a language or languages unrelated to the family that M and N belong to
imperfectly learned a language of that family and adopted the pidgin as
their new mother tongue.
I will adopt the term “Southeast Melanesia” to refer to the part of Melanesia
that lies in Remote Oceania, east of the main Solomons group. It consists
principally of Vanuatu, New Caledonia, the Loyalty Islands and the Fiji
group, and also includes the small Eastern Outer Islands of the Solomons
(including Santa Cruz and its neighbours). More than 140 languages are
spoken in this region: approximately 105 in Vanuatu, 25 in New Caledonia
and the Loyalties, 10 in the Eastern Outer islands and at least three in Fiji
(including Rotuma). A few of these languages are Polynesian Outliers. With
the exception of three languages mentioned below, the rest are generally agreed
to be “Melanesian”.
I will use “Northwest Melanesia” for the part of Melanesia that lies in
Near Oceania and consists principally of the main Solomons group, the
Bismarck Archipelago (New Britain, New Ireland and the Admiralties) and
New Guinea and its small satellite islands and island groups. There are about
300 Melanesian languages in Northwest Melanesia.4 In Northwest Melanesia
there are also some 700 non-Austronesian languages of diverse genetic stocks;
these are often collectively referred as “Papuan” languages.
Languages that are aberrant in one or more of characteristics (i)-(iv)
listed above predominate in certain parts of Southeast Melanesia, namely
in Southern Vanuatu, the Loyalties and New Caledonia and in the Eastern
Outer Islands Province of the Solomons. In these areas some of the aberrant
Andrew Pawley
219
languages exhibit (i) few cognates with other Austronesian languages, and
all exhibit (ii) radical phonological change. However, most are not markedly
atypical in respect of (iii) morphology and syntax. Three languages, those
of the Reefs-Santa Cruz area, have appeared so deviant in respect of criteria
(i)-(iii) that their status as Austronesian has long been in question (Lincoln
1978, Wurm 1978), and François (2006) has recently asked whether the three
Vanikoro languages are really Austronesian. Only in a few cases—chiefly
certain languages of New Caledonia (Grace 1973, 1981, 1992, 1996)—has
(iv) been noted as a very marked characteristic, but there are many regions
of Southeast Melanesia where systematic studies of sound correspondences
among neighbouring languages have not yet been carried out.
Languages at or near the exemplary end of the scale are present in certain
parts of Central and North Vanuatu (such as Nguna of North Efate, Ambae,
Maewo, Raga of North Pentecost, many languages of Espiritu Santo and
Mota of the Banks Islands) and in Fiji (with the partial exception of Rotuman,
which has undergone considerable change in its sound system and, to a lesser
extent, its morphology).
In Northwest Melanesia there is similar variation. Exemplary languages
occupy the eastern half of the main Solomons chain (Florida, Guadalacanal,
Malaita and Makira) but in the Western Solomons and the Bismarck
Archipelago most languages lie between the middle and the aberrant end of
the scale in one or more respects, with exemplary languages in the minority.
The same can be said of New Guinea and its satellites.
The presence of many aberrant Austronesian languages in Southeast
Melanesia is more puzzling than such a presence in Northwest Melanesia.
It is now generally accepted, on archaeological and linguistic grounds, that
Austronesian speakers arrived in Northwest Melanesia just over three millennia
ago and encountered established populations speaking non-Austronesian
languages. 5 In many parts of Northwest Melanesia there was sustained
bilingualism between Austronesian and non-Austronesian communities and
sometimes language shift occurred. Southeast Melanesia is different. It is
generally thought that there were no pre-Austronesian inhabitants of Remote
Oceania. However, Blust (2005) has recently challenged this view.
In this paper I build on sketches of the history of ideas on the origins of
the Melanesian languages that George Grace has included in a number of
publications (e.g., Grace 1959, 1961, 1964, 1971a, 1981). The fullest of these
sketches are in Grace (1959, 1971a), and the later paper takes the story up to
the late 1960s. I will first review the pre-1970 period, during which several
radically different explanations of Melanesian linguistic diversity were in
competition, and will describe how some of these have been ruled out both on
logical grounds and by the weight of evidence. I will then consider research
over the past 35 years or so and assess how much progress has been made in
understanding how the Melanesian languages got to be the way they are.
Figure 1. Map of Oceania, showing the boundary between Near and Remote Oceania.
220
Explaining the Aberrant Austronesian Languages
Andrew Pawley
221
1. ORIGINS OF THE “MELANESIAN” LANGUAGES:
COMPETING HYPOTHESES
1.1 The discovery of the Melanesian languages
The Melanesian languages were late admissions to the Austronesian
family and as recently as the 1960s a few linguists were reluctant to accept
them as full members. Although the data were fragmentary, by the early
1700s European scholars had found evidence of a far flung language family,
already recognised as containing Malay, Javanese and Malagasy (Reland
1708), from comparisons of wordlists of languages of Southeast Asia and
the Pacific Islands brought back by voyagers. Building on groundwork laid
by Joseph Banks, J.R. Forster (1778:188-89) demonstrated the existence of
the Austronesian family (not then called by that name) in a remarkable Table
that showed agreements between Malay and languages from the Philippines
(Tagalog and Kapampangan) and five languages of Polynesia in the numerals
for 1-10 and in forms denoting some 36 other concepts. The Table also
shows that some agreements, though fewer, extended to certain languages
of Melanesia (Tanna and Malekula of Vanuatu), and one language of New
Caledonia for which Forster had data. Although Lorenzo Hervas y Pandero is
sometimes credited with this achievement, the Table that appears in volume 17
of his Idea dell’ Universo (Hervas 1784) probably owes something to Forster.
Indeed, Forster preceded Sir William Jones in anticipating modern ideas about
the development of related languages from a common proto-language.6 Eight
years before Jones’ celebrated statement that Sanskrit, Greek and Latin must
have “sprung from some common source”, Forster commented as follows on
the agreements in his table.
I am … inclined to suppose, that all these dialects preserve several words of
a more ancient language,… which gradually divided into many languages,
now remarkably different. The words therefore of the language of the South
Seas isles, which are similar to others in the Malay tongue, prove clearly
in my opinion, that the Eastern South Sea Islands were originally peopled
from the Indian, or Asiatic Northern isles; and that those lying more to the
Westward, received their first inhabitants from the neighbourhood of New
Guinea. (Forster 1778:190)
Both Forster and Hervas excluded from the Austronesian family the
languages of Melanesia, for which only the most fragmentary data were
available. Forster identifies two main “races” in the South Seas and, like others
of his time, was inclined to think that linguistic divisions would consistently
correlate with those of race.
The first wordlists from the South Pacific were collected in 1568 during
Mendaña’s expedition and consisted of 38 words from S.E. Santa Isabel in the
Central Solomons. In 1605 Quiros gathered short vocabularies from Sikaiana
222
Explaining the Aberrant Austronesian Languages
(a Polynesian Outlier north of the Solomons) and Espiritu Santo in Vanuatu. In
1616 the Dutch voyagers Schouten and Le Maire recorded vocabularies from
Niuatoputapu, between Tonga and Samoa, and Futuna, west of Samoa, and three
vocabularies from islands off New Guinea and New Ireland. Over a century
and a half later the voyages of Cook and d’Entrecasteaux brought back short
vocabularies from New Caledonia and Vanuatu as well as more extensive ones
for several Polynesian languages. By the late 18th century the existence of a
far-flung Polynesian group was well known in Europe and its relationship to the
family that contains Malay, Javanese and Malagasy was accepted. It was thought
that the Polynesian branch stemmed from the migration of wavy-haired, brownskinned seafarers, essentially Malay type peoples, from Southeast Asia.
Around the middle of the 19th century missionary scholars began to study
some of the languages of Melanesia in systematic fashion. By the 1840s
and 1850s such work was under way in parts of Fiji, Vanuatu, the Loyalties,
New Caledonia and the Eastern Solomons. It was soon evident that many of
these languages showed a family resemblance to those of Polynesia and the
Indo-Malaysian region. It was probably the German linguist Hans Conon
von der Gabelentz who first used the term “Melanesian languages”, in the
first of his two volumes of grammatical sketches of languages of Melanesia
(Gabelentz 1861-1873). Gabelenz distinguished between “Melanesian” and
“Polynesian” languages but assigned both to the “Malay” family.
Many 19th century observers were perplexed by the fact that languages
resembling those of Polynesia and Indonesia were spoken by dark, frizzyhaired people. A common view was that the Melanesian peoples must have
borrowed Austronesian elements from their neighbours to the west and/or the
east. The term “Papuan” to refer to people of the South Pacific whose physical
type was distinct from both Malays and Polynesians had been introduced by
William Marsden in the 1830s. During the 1850s the naturalist A.R. Wallace
collected vocabularies in the eastern part of the Indo-Malaysian Archipelago.
When he published these (Wallace 1869) he proposed a distinction between
“Malay” and “Papuan” languages in that region, a distinction that was to be
echoed in Melanesia a few decades later.
1.2 Codrington’s The Melanesian Languages
By the 1880s there was enough material on the languages of Southeast
Melanesia for Bishop Robert Codrington of the Melanesian Mission to publish
the first major comparative linguistic study of this region, together with sketch
grammars of 34 languages. In The Melanesian Languages Codrington (1885)
sought to prove conclusively that the “Melanesian” languages are Austronesian.
He was clearly exasperated by prevalent race-based interpretations of the
linguistic facts and commented on these with heavy irony.
Andrew Pawley 223
The Melanesian people have the misfortune to be black, to be much darker,
at least, than either Malays or Polynesians; and because they are black it is
presumed that their original language cannot be of the same family with that
spoken by their brown neighbours; that where their language has a general
resemblance to that of their neighbours they must have cast off their own and
taken another in the lump, and that where the resemblance is not conspicuously
apparent they must have borrowed words and expressions in commercial or
other intercourse. (Codrington 1885:12)
Codrington showed that almost all the Melanesian languages in his sample
share many everyday words and a good deal of morphology (the grammar of
word formation) with well-established members of the Austronesian family.
Furthermore the Melanesian languages share some features that distinguish
them from those of the Indo-Malaysian archipelago and Polynesia. These
typically “Melanesian” features included several in morphology: in nouns, a
distinction between alienable and inalienable possession and often between
two or three kinds of alienable possession (food versus drink versus general);
and in verbs, a transitive suffix ending in -i or -aki.
The view of the Melanesian Languages here proposed is, in the first place, that
they are homogenous; and secondly, that they belong to a common stock with
the Ocean [i.e., Austronesian] tongues generally— those of the Indian [i.e., IndoMalaysian] Archipelago and of Polynesia. The view which is opposed is one
which would make the Melanesian stock of languages originally distinct from
that to which Malayan and Polynesian languages belong, and would pronounce
all that is found in Melanesian languages common with Malay and Polynesian
to be borrowed from these tongues, or due to influence received from them.
In opposition to this latter view, it is by no means denied that the Melanesian
languages have borrowed from those of the Indian or Malay Archipelago on
the one side, and from those of the Eastern Pacific [i.e., Polynesian] on the
other, or that they have been influenced in various ways; allowing this, what
is maintained is, that whatever has been introduced has been brought from
languages of a kindred, not a distinct stock. (Codrington 1885:10)
By the close of the 19th century there were further developments. Some
scholars now argued that the Polynesian group had as their closest relatives
certain languages of Melanesia: Kern (1886) proposed Fijian, while Schmidt
(1899a, 1899b) favoured Gela, Bugotu and Vaturanga of the Central
Solomons. However, it was a very different view that was to gain more
general acceptance, if not among linguists, then among anthropologists and
others writing on the history of Pacific Island peoples and to hold sway into
the second half of the 20th century.
224
Explaining the Aberrant Austronesian Languages
1.3 Ray’s theory of the pidgin origins of Melanesian languages, or, How
some Papuan languages became Austronesian by degrees
Sidney Ray, all his working life a schoolteacher in the East End of London,
was inspired by Codrington’s 1885 book and in 1886 began to accumulate
a vast body of data on languages of Melanesia. Although Ray carried out
fieldwork in Southeast Papua with the British ethnological expedition to
the Torres Straits in 1898, the main sources of data for the many works he
published over the next 40 years were Bible translations, provided by the
British and Foreign Bible Society. In the early 1890s Ray pointed out that
some of the languages of British New Guinea and the Solomon Islands did not
appear to be Austronesian at all (Ray 1892-93). Following Wallace’s lead, he
adopted “Papuan” as a name for this residual category of languages. Around
the same time Pater Wilhelm Schmidt (1899a) observed that “Papuan” as
well as “Austronesian” languages were present on and near the north coast
of the New Guinea mainland and in New Britain.
Ray’s major synthesis, A Comparative Study of the Melanesian Island
Languages, was published in 1926. It focused on the Solomons, the Loyalties
and Vanuatu, largely overlapping with the region studied by Codrington but
extending to the Western Solomons. In this 600 page book Ray presented
grammatical sketches of more than 50 languages, listed reflexes of ProtoIndonesian words as reconstructed by Renward Brandstetter, and attempted
to trace the reflexes of Proto-Indonesian sounds in each language.
Ray’s conclusions were essentially those that Codrington had been at
pains to disprove. The Melanesian languages are diverse in appearance, Ray
said, because they are not truly Austronesian. Instead they derive from pidgin
languages, which developed when Papuan-speaking communities imperfectly
learnt the Austronesian languages of migrants from the Indo-Malaysian
Archipelago, a process that happened independently many times. We should let
Ray speak for himself (bearing in mind that by “IN [Indonesian] language” Ray
means any Austronesian language from west of Melanesia, e.g., from Malaya,
the Indo-Malaysian Archipelago, the Philippines, Taiwan and Madagascar).
The great variation in the extent to which the Melanesian islanders have changed
IN words seems to suggest that these words were introduced by colonists from
Indonesia, who effected a settlement on the smaller islands, imposing part of
their speech upon the natives, and that this mixed speech influenced the native
languages with which it came in contact, these latter adopting some of the IN
modified speech, but changing it according to their own style of pronunciation.
An IN word, for example, like manuk becoming manu in the colony, but man,
mon, men, min, in districts away from it. (595-96)
The IN words found in Melanesia have the characteristics of a pidgin-tongue.
They can be no longer referred, except in rare cases, to any one original IN
tongue, and are on a par with the modern pidgin of the Pacific where the co-
Andrew Pawley 225
called English has such words as “savvy,” “pickaninny,” and “wewe”. (597)
The very large number of words which cannot be referred to an IN source
cannot be shown to have (except to a very slight extent) any community of
origin. This diversity, which is characteristic of the Papuan languages of New
Guinea, New Britain and the Solomon Islands, suggests that the Melanesian
languages were originally of variant stocks, and that their apparent uniformity
has been brought about by the influx of IN words and idioms. (596)
However, Codrington had made his case on the basis of agreements in
grammar (specifically morphology) rather than lexicon. Ray acknowledged
these grammatical agreements in the following terms, but interpreted them
differently.
In grammar the languages depart widely from a common standard, yet there is in
each area a use of the broad outstanding features of IN grammar. Such are:
1. A general agreement in the numerals from ‘one’ to ‘five’ where the system
is quinary, and in the higher numbers when decimal.
2. A general agreement in the basic form of the pronouns.
3. The use of a short pronoun with the verb.
4. A tendency in some languages to use the verb without a pronoun in the
third person singular.
5. The persistence here and there of ready-made IN words, such as
matakut, afraid, maturu, sleep, mauri, live, mavat, heavy, etc., with no
corresponding simple form.
6. The use of a suffixed pronoun with nouns, but only to parts of the body,
or ‘definite property of me, thee,’ etc. Other possessions being ‘thing of
mine, thine,’ etc.
7. The excessive and exclusive use of the IN prepositions i, ni and ki. Cf.
‘belong’ and ‘along’ in pidgin-English. (Ray 1926:597)
Ray explained such grammatical agreements as stemming from numerous
episodes of pidginisation, in which speakers of Papuan languages in different parts
of Melanesia independently acquired a simplified but broadly similar version of
Indonesian grammar. There are several obvious objections to his theory:
(i) Codrington, the more astute historical linguist, had already pointed
out that some of the Melanesian morphological agreements have no
counterparts in Indonesian and Polynesian languages, so they could not
have been derived from them by simplification. And it is very unlikely
that certain striking agreements in details of morphology would have
developed independently or from Papuan sources, if as Ray insisted,
Papuan languages were themselves of diverse stocks.
226
Explaining the Aberrant Austronesian Languages
(ii) Not a single Papuan language survives in the core region of Codrington’s
and Ray’s studies: the Southeast Solomons, Vanuatu, Loyalties and New
Caledonia. Ray’s theory thus requires every Papuan language in this large
region to have become extinct—not impossible, perhaps, but a pretty
uneconomical hypothesis.
(iii) No place is given to that most natural of linguistic processes, change
resulting from system-internal pressures. Ray attributes practically all
departures from the Indonesian prototype exhibited by the Melanesian
languages to contact between speakers of Papuan and Austronesian
languages. But no direct evidence of a Papuan substratum in the
Melanesian languages is cited. Indeed, as no Papuan languages are found
in Southeast Melanesia the only linguistic argument for saying that Papuan
languages were ever spoken there was the observation that some of the
Melanesian languages shared relatively few cognates with one another
or with Polynesian and Indonesian languages and the belief that they had
“simplified” the sound system and grammar of Proto Indonesian.
What led Ray to overlook these weighty objections? He points out that the
discovery in the 1890s that there are non-Austronesian languages in various
parts of Melanesia required a revision of the idea that the languages of
Melanesia are all “related as descendants of an original mother tongue” (Ray
1926:25). But as the only clear cases of non-Austronesian languages were
confined to Northwest Melanesia that does not explain why Ray should reject
Codrington’s view of the origins of the languages of Southeast Melanesia in
such a sweeping manner.
It seems that the difference between Codrington’s and Ray’s interpretations
had little to do with the range of data at their disposal. It owed much more
to their respective mindsets and biases. Codrington saw race as irrelevant to
the question of linguistic affinities. He looked at his Melanesian glass with
the eye of a classical comparative linguist and saw that it was half full of
resemblances that provided good evidence for gradual and systematic change
from a common Austronesian ancestor. He began with the more exemplary
languages and found he could make sense of them according to the standard
family tree theory. He then assumed, as a matter of faith, that problems in
accounting for the more deviant languages would somehow be resolved
without disturbing his main hypothesis.
Ray, on the other hand, looked at the same glass and focused not on the
resemblances but on the differences. He took the more aberrant languages
to be the norm and saw them as incapable of being explained in terms
of gradual and systematic change. He then assumed, as a matter of faith
that the explanation he proposed to make sense of the aberrant languages
Andrew Pawley
227
would also do for those that show a higher degree of likeness to the wellestablished Austronesian languages of Indonesia. Why did Ray take the
aberrant languages to be the norm? Almost certainly he was influenced by
the idea that there was an original correlation between language family and
racial type. Melanesians were seen to be of “Papuan” physical type. Some
Melanesian peoples speak Papuan languages and therefore all Melanesians
once spoke Papuan languages.7
1.4. Dempwolff defines the Oceanic subgroup
Soon after Ray’s book appeared there was a major advance in Austronesian
historical linguistics. Otto Dempwolff had spent several years in German
New Guinea around the turn of the century, combining linguistic research
with work as a medical doctor. Building on a series of papers he wrote in the
1920s, and on the work of earlier scholars such as Brandstetter, Dempwolff
(1934-38) published a three volume work that provided by far the most
systematic reconstruction of the Proto-Austronesian (PAn) sound system
to date along with more than 2000 reconstructed roots. The reconstructions
he attributed to “Proto Austronesian” are nowadays equated with the later
inter-stage known as “Proto Malayo-Polynesian” (PMP), ancestral to all
Austronesian languages outside of Taiwan; the reason is that Dempwolff’s
sample excluded the Austronesian languages of Taiwan, which are now known
to be genetically very diverse (see 1.7).
An important by-product of Dempwolff’s comparative phonological
work was proof of the subgroup that we now call “Oceanic”.8 His sample
of languages from Melanesia showed largely regular correspondences with
PMP consonants and vowels.9 They also showed something even more
striking: in vocabulary inherited from PMP the Melanesian languages share
a sizeable set of common innovations to the PMP sound system, innovations
that are not present in Austronesian languages found west of New Guinea.
Furthermore, the Polynesian and Micronesian languages (except Chamorro
and Palauan) that he examined have undergone the same innovations. These
shared innovations can thus be attributed to a common inter-stage, Proto
Oceanic (POc), ancestral to just the Melanesian, Polynesian and Micronesian
languages. However, the exact position of the western boundary of Oceanic
in New Guinea was unclear in Dempwolff’s time.
The main phonological innovations attributable to POc are as follows.
For convenience I refer here to contemporary reconstructions, which make
various modifications to Dempwolff’s reconstructions and orthography. In
the consonants, there were several mergers (where contrasts between certain
pairs of PMP consonants were lost) in POc. PMP *p and *b merged, as did *k
and *g, *d and *r, and *s and *Z. PMP *h was lost. Several new consonants
228
Explaining the Aberrant Austronesian Languages
were added: the labiovelars *bw, *pw and *mw. In the vowels, PMP *e and
*aw merged as *o, *i and *uy merged as *i. There were some other rather
complicated developments that need not detain us here (for a fuller account
see Lynch et al. 2002:63-67).
Dempwolff found that more than 600 of the 2000 or so PMP roots he
reconstructed have reflexes in Melanesian languages. (Today’s figures for
both PMP etyma and reflexes in Melanesian languages are much higher.)
Dempwolff’s proofs of an Oceanic subgroup should have greatly narrowed
the debate over the history of the Melanesian languages. It is incompatible
with two of the five major types of explanation for the aberrant languages,
namely genealogical diversity—the idea that the aberrant languages are only
distantly related to the exemplary ones, and their separate origins as pidgins.
The aberrant languages do not belong to different subgroups of Austronesian
from their exemplary sisters. It is just that they have changed faster. And
if the Melanesian languages all fall into an Oceanic subgroup apart from
all “Indonesian” languages, it follows by definition that the Melanesian
languages could not be descended from many different pidgin languages
that arose independently as a result of various Papuan languages taking on
layers of Indonesian borrowings.
However, the matter was not yet settled. Over the next few decades the
Oceanic hypothesis continued to come under indirect attack from certain
quarters. I say “indirect” because its critics did not try to demonstrate flaws
in Dempwolff’s evidence (even though there were some). Instead they chose
to address quite different kinds of evidence and to use different conceptual
frameworks.
1.5 Capell’s support for Ray
In the 1930s Ray’s baton was taken up by the Australian clergyman-linguist
Arthur Capell, the leading Australian linguist of the first 60 years of the 20th
century. Like Ray, Capell was a dyed-in-the-wool diffusionist and he was
never fully committed to the logic of the family-tree model and the method
of determining subgroups by shared innovations.
Capell’s leanings were made evident in a short paper he published in 1933
where he compared the structure of “Melanesian” and “Papuan” languages.
He begins by saying that the paper is “in no sense a dissertation of the vexed
and undecided question of Melanesian origins” (Capell 1933:418), but in the
end is unable to resist putting forward some fairly firm views. Although “the
general structure of Melanesian grammar is fairly uniform from New Guinea
to Fiji, and is very similar to that of the Polynesian languages”, in regard to
vocabulary the “common element is remarkably small. Ray…lists seventy
words as being common throughout Melanesia, which can be traced back to
Andrew Pawley
229
Indonesia. The addition of others very common but not clearly Indonesian
might bring the list up to a hundred or so [but] all the remainder appears to
be ‘aboriginal’…” (432-33). Capell concludes that:
[There is no] real doubt that the “Papuan” races represent the aboriginal
populations of the islands, and we have to imagine wave after wave of
migration from the west, making impact on these “Papuan” populations and
modifying their languages and customs in different ways and in different
places. (431)
If we regard the original linguistic condition of Oceania as that of many
practically unrelated tongues spoken by tribes having very little contact
with each other, then it becomes obvious that the present common element
must have been introduced and superimposed….The Indonesian has simply
impinged on the original, different words being taken in different places,… The
Indonesian element in Melanesia… is definitely “pidgin”, i.e. constructions,
grammatical elements, etc. have frequently been taken over with very little
regard for their original meaning. (432-33)
Capell’s ideas were developed further in his doctoral thesis, completed
at the University of London in 1938 (Capell 1943). This consisted of a
detailed comparative study of the lexicon of the Austronesian languages of
Southeast Papua, which he divided into 11 different groups. To account for
the distribution of particular cognate sets in these groups and in the IndoMalaysian region he proposed three separate migrations into Southeast
Papua from the west, one from Borneo, another from Central Sulawesi
and the Philippines, and a third from Java-Sumatra via the Philippines. He
assumed that if a word in Southeast Papua has a “western” cognate only in,
say, Borneo, it was evidence for a migration from Borneo. Capell made use
of the phonological and lexical reconstructions in Dempwolff (1934-38) but
did not accept the sub-grouping indicated by these reconstructions.
Later work by the German businessman-linguist, Wilhelm Milke (1961),
showed that Capell’s three migration theory had no significant statistical basis.
Nevertheless, Capell himself continued to support a version of the Ray theory,
albeit in more guarded form (Capell 1962, 1971). And in popular treatments
of Pacific culture history written by anthropologists and archaeologists until
the 1970s this theory remained dominant, or was given equal status with the
Oceanic hypothesis and a third hypothesis to which we now turn.
1.6 Dyen’s (1965) lexicostatistical classification: The Melanesian
homeland hypothesis
A second indirect assault on the Oceanic hypothesis came from an
unexpected quarter. In the mid 1950s Isidore Dyen, the leading Austronesian
230
Explaining the Aberrant Austronesian Languages
historical linguist of the time, who until then had worked very much within
the framework constructed by Dempwolff, became interested in applying to
the Austronesian family a new method of sub-grouping: lexicostatistics.
Lexicostatistics can be described as a quick and dirty quantitative method
of arriving at a family tree classification. The idea is that a sub-grouping can
be arrived at simply by counting the number of cognates that sister languages
share with each other in a sample of “basic” vocabulary, usually consisting
of 100 or 200 words. Basic vocabulary items refer to putatively universal
concepts, such as body parts, constant features of the physical environment
like sun, moon, water and rain, activities like eating, drinking and sleeping,
and certain personal pronouns and kinship relations. It is generally agreed
that words for these concepts are less likely to be borrowed than “cultural”
vocabulary and that they tend to be replaced rather slowly (where “replaced”
means that a word is either lost or ceases to be the most common word for
the meaning in question). A small number of control studies of (mainly IndoEuropean) languages with long written histories indicated that they all show
a replacement rate within two or three points of 19.5 percent per millennium
on the 200 item list. That is, an average of about 80 percent of the words on
the list were retained over 1000 years. These control studies were also the
basis for glottochronology, a method by which approximate absolute dates
for language splits are derived from lexicostatistical percentages.
Dyen (1965) compared standardised 196 item wordlists for some 245
different Austronesian languages—a sizeable sample although we now know
it amounts to less than a quarter of the total number of languages in the
family. Some 70 of the languages were Melanesian. There were weaknesses
in sampling, data and decision-making on cognation, nonetheless the study
showed beyond reasonable doubt that, within the Austronesian-speaking area,
Melanesia contains by far the greatest number of languages showing very low
cognate counts with all languages other than their immediate neighbours. He
argued that, taking the lexicostatistical evidence at its face value, the most
reasonable interpretation of the matrix of cognate percentages for the 245
languages is that they divide into 40 first order branches of Austronesian, of
which no fewer than 31 are located in Melanesia. Many of these are single
member branches.
This finding should not have surprised anyone familiar with the previous
century of debate on the origins of the Melanesian languages. What was
surprising is Dyen’s assessment of the implications. He concluded that
the lexicostatistical evidence amounts to a strong case that Melanesia is
the homeland—the primary dispersal centre—of the entire Austronesian
family. Within this region, the New Britain area, as the region of greatest
lexicostatistical diversity, is the most likely homeland. Having made this
Andrew Pawley
231
observation Dyen noted, albeit very briefly, that this hypothesis conflicts with
that of an Oceanic sub-group based on shared innovations in phonology. He
pointed to three other regions as alternative but weaker candidates for the
homeland, on the grounds of their considerable lexicostatistical diversity,
namely (i) Taiwan, (ii) Sumatra and the islands off its west coast, (iii) West
New Guinea and the eastern part of the Indonesian archipelago.
The Melanesian homeland hypothesis had a spectacular but short-lived
career. The hypothesis made sense only on the assumption that most of the
languages in the study have replaced their basic lexicon at roughly constant
rates. To most Austronesianists this seemed very unlikely for a number of
reasons—one being that Dyen’s lexicostatistical classification conflicted
sharply with the well-supported Oceanic hypothesis. In his review of Dyen
(1965), Grace (1966) suggested that the exceptional lexical diversity of
Melanesian languages is most likely due to many languages there having
relatively low retention rates. Grace (1967) demonstrated variability in
retention rates for 20 of the most stable items in the basic vocabulary list
in six Oceanic languages. Some 15 years later Blust provided proof of such
variability on a much larger scale (see 1.8 below).
1.7 Consolidating the Oceanic hypothesis
In the 1950s and 1960s, as more university departments began to employ
scholars trained in modern linguistics, Austronesian linguistics became
professionalised. For the first time there were considerable numbers of trained
linguists and their students writing grammars and dictionaries of Austronesian
languages. The new generation of Austronesianists included some who did
historical linguistics, at first just a handful, gradually increasing to several
dozen. The journal Oceanic Linguistics was established in 1962. The first
international conference on Austronesian linguistics was held in Honolulu
in 1974 and became a regular event.
Among Oceanicists the main focus of historical work in the 1950s, 60s
and early 1970s was on sub-grouping and on phonological, lexical and
grammatical reconstruction. Reconstructive work relied heavily on exemplary
languages. There were good reasons for this. In the first place, most of the
best-described Oceanic languages belonged to the exemplary category. In the
second place, conservative daughter-languages are much better witnesses than
aberrant daughter-languages for reconstructing the proto-language. Starting
with ill-behaved daughter-languages is a poor strategy, likely to bring more
headaches than solutions. Once the history of the conservative languages is
reasonably well understood this provides established points of reference for
forays into more difficult territory.
232
Explaining the Aberrant Austronesian Languages
One important outcome of this work was the consolidation of the Oceanic
hypothesis. Additional innovations defining an Oceanic subgroup were found,
mainly in the domain of morphology and morphophonology (e.g., Blust 1977,
Lichtenberk 1985, Lynch 1996, Pawley 1973). Numerous refinements to the
sound system of POc have been made (Blust 1978, 1981a; Haudricourt 1971;
Lichtenberk 1978; Milke 1968; Lynch 2000; Lynch, Ross and Crowley 2002;
Ross 1988, 1989; Wolff 1974). The body of POc lexical reconstructions has
grown to upwards of 1500 (the most detailed account is in Ross et al. (1998,
2003, in prep.), although Blust 1995a contains more than 4000 PMP lexical
reconstructions with supporting cognate sets many of which have Oceanic
reflexes). Extensive sets of lexical reconstructions for various lower-order
proto-languages within Oceanic stages have also been made, including
Proto Polynesian (Biggs and Clark n.d.), Proto Micronesian (Bender et al.
2003), Proto Central-North Vanuatu (Clark n.d.) and Proto Southern Vanuatu
(Lynch 2001).
The upper sections of the Austronesian family tree have been quite
drastically revised, chiefly in a series of papers by Blust between 1977 and
1999, with evidence accumulating that Oceanic is no higher than a fourthorder branch of Austronesian (Blust 1995b), as indicated in Figure 2.
Figure 2. An Austronesian family tree (after Blust 1995b).
Figure 3. Distribution of the Austronesian family and its major subgroups.
Andrew Pawley
233
234
Explaining the Aberrant Austronesian Languages
The western boundary of Oceanic in New Guinea was determined as lying
between 136 and 138 degrees East (see Figure 3). Blust (1978) showed that
the Austronesian languages of Cenderawasih (Geelvink) Bay and South
Halmahera are not Oceanic but form a subgroup that is almost certainly the
closest relative of Oceanic. Grace (1971b) and Ross (1996a) showed that the
languages of the Sarmi coast and Jayapura, to the east of Cenderawasih Bay,
have undergone at least some of the sound changes diagnostic of Oceanic
(other changes are indeterminate).
1.8 Blust’s proof of variable retention rates
Grace’s point about variable retention rates was proved beyond reasonable
doubt in an ingenious study by Robert Blust. Using as his baseline Proto
Malayo-Polynesian (PMP) lexical reconstructions representing some 200
basic concepts, he compared 55 Malayo-Polynesian languages and showed
that they vary greatly in their rates of lexical retention (Blust 1981b). Later
he extended the sample of languages to 230 (Blust 1999). The lexically most
conservative language in his sample was Malay, which has kept 58 percent
of the PMP words. The most conservative Oceanic languages, such as Fijian
and Samoan, have kept around 40 percent, similar to the average for most
Western Malayo-Polynesian and Central Malayo-Polynesian languages. The
average number of retentions for the Oceanic languages is, however, much
lower: 23.6 percent, chiefly owing to low scores among certain languages
of Melanesia. The most innovative language was Kaulong of New Britain,
which retains only 5 percent. A number of other languages of Melanesia score
between 7 and 12 percent.
1.9 Estimating absolute rates of lexical replacement: Enter archaeology
Blust’s study of variable retention rates allows the notions “lexically wellbehaved” or “lexically aberrant” to be quantified in terms of the proportions
of basic vocabulary items retained by sister languages since the breakup of
a common ancestor. But other discoveries now allow us to go further and
speak with some confidence about the absolute rates of lexical replacement
in particular Malayo-Polynesian languages.
Until recently linguists had little idea of the precise dates and directions
of the early stages of the dispersal of Austronesian languages. However,
archaeological and linguistic work over the past few decades has changed
that. The spread of the Lapita culture and its forebears in Island Southeast
Asia is strongly associated with the initial spread of Austronesian languages
(Bellwood 1997; Green 2003; Kirch 2000; Pawley 2002, 2003; Shutler and
Marck 1975; Spriggs 1997).
It now appears that the first Neolithic culture to enter Island Southeast
Asia came into Taiwan from southeast China about 5,500 years ago. This
Andrew Pawley
235
culture was characterised by sedentary settlements with substantial wooden
houses, pottery-making, cultivation of rice, domesticated pigs, chickens
and dogs, and a distinctive array of tool and ornament types. After some
1500 years of local diversification in Taiwan, bearers of one of the Taiwan
regional Neolithic cultures settled the Batanes Islands and Luzon around 4000
years ago (Bellwood and Dizon 2005, in press). From there variants of this
culture spread rapidly to various parts of the Philippines and Indonesia and
to the Marianas and Northwest Melanesia. By 3400-3300 B.P. one variant
had reached the Bismarck Archipelago, in the manifestation known as the
Lapita culture. Between 3200 and 3000 B.P. bearers of the Lapita culture
colonised Remote Oceania, settling in the Santa Cruz group, Vanuatu and
New Caledonia. By 3000-2900 B.P. they had reached Fiji and Tonga.
Although humans colonised New Guinea, New Britain and New Ireland
by 40,000 years ago and reached Bougainville some 30,000 years ago (Kirch
2000, Specht 2005, Spriggs 1997), there is no archaeological evidence of
pre-Lapita occupation of any islands in Remote Oceania. Evidently the preLapita inhabitants of Near Oceania lacked the sailing capabilities and the food
production techniques to establish viable settlements in Remote Oceania.
The break-up of Proto Malayo-Polynesian can be placed at around 4000
B.P.—the time when the Taiwan Neolithic spread into the Batanes Islands. and
Luzon—and the breakup of POc at no later than about 3200-3100 B.P.—the
time when the Lapita Neolithic spreads into Remote Oceania. We can use the
4000 B.P. date as a baseline for estimating retention (or replacement) rates
of the Malayo-Polynesian languages in Blust’s sample. Table 3 shows what
would happen over 4000 years if the average rate of lexical replacement per
1000 years were 10, 20, 30, 40 and 50 percent respectively. The lower figures
in each column show the percentage of original lexicon retained.
Table 3: Percentages of basic vocabulary retained at different replacement rates
over four millennia.
Replacement rates
Millennia
10%
20%
30%
40%
50%
1
90
80
70
60
50
2
81
64
49
34
25
3
73
51
34
24
12
4
65
41
24
14
6
236
Explaining the Aberrant Austronesian Languages
By these calculations, the most innovative language in Blust’s sample,
Kaulong, has replaced its basic lexicon at an average rate of about 50 percent
per millennium, whereas the rate in Fijian and Samoan is about 20 percent.
The replacement rate in the most conservative MP language, Malay, is only
about 12 percent per millennium. The differences between the extremes in
rate per millennium are considerable but it is their cumulative effect over
time that is more striking.
2. PROGRESS SINCE 1970 IN UNDERSTANDING WHY SOME OCEANIC
LANGUAGES OF MELANESIA HAVE CHANGED MUCH MORE THAN OTHERS
2.1 Competing explanations
Showing that languages have changed their lexicons (or other components)
at variable rates is one thing. Providing a satisfactory explanation for extreme
differences in rates of change is another. I turn now to an assessment of
progress since about 1970 in understanding how some Oceanic languages
of Melanesia came to be aberrant.
It was noted earlier that in the 1950s, 1960s and early 1970s the bulk of
historical work in Oceanic used exemplary languages as key witnesses, for
strategic reasons. During the 1970s, with better descriptive data becoming
available for a wider range of languages and with a fairly secure framework
of POc reconstructions in place, more scholars were emboldened to begin
working out the history of some of the aberrant languages.
Five main hypotheses had been proposed to explain the aberrant Melanesian
languages: (a) the extreme pidginisation hypothesis proposed by Ray and Capell,
which holds that all Melanesian languages stem from pidgins, created when
Papuan-speaking communities adopted as their mother tongue an imperfectly
acquired Austronesian language, (b) deep genealogical diversity, (c) extensive
change due to organic (language-internal) processes, (d) extensive change due
to community-internal social processes, and (e) extensive borrowing between
Oceanic and Papuan languages. Of the five, only three remained viable. By
the 1970s the consensus among Austronesianists was that (a) and (b) could be
ruled out, leaving (c), (d) and (e) as the most likely possibilities.
In Northwest Melanesia (e) has clearly been a major factor and I will return
to this in 2.4. But for the Southeast Melanesian region, with the exception of the
Santa Cruz area, (e) looked to be unlikely because there was no direct evidence
that non-Austronesian languages had ever been spoken in that region.
For most of Southeast Melanesia that left (c) and (d). Now it is the natural
bent of historical linguists to seek to explain change in terms of organic (system
internal) processes, as described under (c) in the first section of this paper, and
for most of those linguists who worked on aberrant languages of Southeast
Andrew Pawley
237
Melanesia in the 1970s the question was: how much of the change undergone
by these languages can be satisfactorily attributed to organic change?
To talk usefully about how types of organic change come about it is
necessary to distinguish between different subsystems of a language. The
import of these contrasts has to do with what factors trigger changes, how
much impact a change to one part of the system has on other parts and on
how particular kinds of changes spread through the speech community.
One important distinction is between systems that people are able to
consciously manipulate with some ease and systems that operate in a very
mechanical way, largely below the level of consciousness. Systems where
the units and relations are strongly conceptual (representing concepts with
real world associations) and where the units are isolable fall into the first
category. Broadly speaking, lexical items (roughly, words and conventional
phrases) are the part of language that people can consciously manipulate
with most ease. Lexical units are therefore fairly vulnerable to influence by
social factors. However, insofar as the lexicon is a large, open system with
many relatively independent subsystems, change in one part of the lexicon
has little impact on other parts.
Systems that are weakly conceptual or a-conceptual (the units and their
relations have no meaning in themselves) and where the boundaries of the units
are not very transparent generally operate below the level of consciousness.
Phonological systems fall into this category and for the most part speakers
do not consciously manipulate the units. Such systems can therefore be very
stable. However, because they are tight-knit systems, with interdependent
parts, they can also change dramatically in a fairly short time, when one or
two small systemic changes provide the preconditions for further changes
which in turn may “bring the house down”, so to speak. The far-reaching
effects of regular consonant-vowel metathesis on vowel systems (Blevins
and Garret 1998), such as has occurred in Rotuman, Kwara‘ae and Ririo (see
2.2), provide one class of examples, and the sequence of changes exhibited
by Southern Vanuatu languages (also see 2.2) provide another.
One is tempted to say that grammar lies between these two extremes.
But in fact the heading “grammar” subsumes a cluster of diverse systems
that are relatively independent of one another. Particular morphological
paradigms—ways of forming words by sets of inflectional and derivational
affixes—tend to be closed, tight-knit and low on conceptual content.
Morphological subsystems may be extremely stable over long periods of
time but may also be transformed quite rapidly as a result of sound change
or other system-internal changes. There are other systems of grammatical
markers that are intermediate between morphology and lexicon. Syntax, the
rules for combining words into larger constructions, is made up of some
238
Explaining the Aberrant Austronesian Languages
rather highly structured subsystems that are very largely a-conceptual, but
some elements of syntax, such as word order, are more open to conscious
manipulation than morphology.
Of course, all innovations begin with individual speakers. Whether an
innovation is nipped in the bud or spreads through an entire speech community
(or some part of it) is determined by social and demographic factors. Oceanic
linguists have seldom undertaken case studies of the spread of innovations
across speech communities, recording the kinds of information from which
comparative linguists might draw generalisations. Geraghty’s (1983) work
on Fiji remains by far the best study of dialect geography combined with a
historical perspective.
In recent decades considerable progress has been made in reconstructing
the sequence of changes undergone by some of the aberrant languages. My
comments on this work will necessarily be very sketchy. In the last 35 years
upwards of 500 papers and books have been published on Austronesian
historical linguistics—indeed, more work has been done in this period than
in the previous 200 years. Around half of these works have dealt with issues
in Oceanic historical linguistics.
2.2 Phonological change
POc is reconstructed as having a rather simple syllable and root structure.
Syllables consisted of a vowel alone or a consonant plus a vowel, with wordfinal syllables able to have a final consonant. Roots could consist of one,
two, three and occasionally more than three syllables. It follows from syllable
structure constraints that no phonemic consonant clusters were allowed in
roots or complex words, e.g., disyllabic roots had the shape (C)V(C)V(C)
and trisyllables (C)V(V)V(C)V(C). Twenty three consonants and five vowels
(*a, *e *i *o *u) are reconstructed (Lynch et al. 2002).
Many languages of Southeast Melanesia have drastically modified the POc
sound system. For example, some New Caledonian languages have developed
very complex vowel systems, some with more than 20 contrasting vowels,
including contrasts between nasal and oral vowels (Gordon and Maddieson
1999, Haudricourt 1971, Rivierre 1973), and tone contrasts. Iaai, in the
Loyalties, has developed 37 consonants (Ozanne-Rivierre 1976). In some
languages the inventory of consonants had been considerably reduced, as a
result of losses and mergers.
We certainly know a great deal more about how certain languages of
Southeast Melanesia came to be phonologically aberrant than was the case
in 1970. A linguistic survey of Vanuatu by Tryon (1976) yielded tables
listing the most common reflexes of POc consonants and vowels of 179
languages and dialects. Tryon and Hackman (1983) did the same for nearly
Andrew Pawley
239
Table 4: Probable sequence of sound changes to some POc words in Sye,
Lenakel and Anejom (adapted from Lynch 2001:105-6).
1. Sye
POc
*e tama-ña
*kulit
*makubu-ña
Pre-PSV
e-taˈma-na
na-kuˈliti
makuˈbu-na
(Pre-deletion rules)
e-təˈma-na
no-kuˈlisi
mokuˈbu-na
MEDIAL VOWEL DELETION
e-tˈma-na
no-kˈlisi
mokˈbu-na
FINAL VOWEL DELETION
ˈe-tma-n
no-kˈlis
ˈmokbu-n
(Other rules)
ˈetme-n
noˈɣleh
ˈmoɣpo-n
‘his father’
‘skin’
‘his grandchild’
POc
*panako
*na lima-ña
+
Pre-PSV
a-paˈnako
na-liˈma-na
na-baˈyani
2. Lenakel
na bayan
(Pre-deletion rules)
a-pəˈnako
ne-liˈma-na
na-bəˈyani
MEDIAL VOWEL DELETION
a-pˈnako
ne-lˈma-na
na-bˈyani
FINAL VOWEL DELETION
ˈa-pnak
ˈne-lma-n
ˈna-byan
(Other rules)
ˈəvnak
ˈnelmə-n
ˈnəpien
‘steal’
‘his hand’
‘bait’
POc
*keli
*na lima-ña
+
Pre-PSV
a-keˈli-i
na-liˈma-na
na-baˈyani
(Pre-deletion rules)
a-keˈji-i
ne-jiˈma-na
ne-bəˈyañi
MEDIAL VOWEL DELETION
a-kˈji-i
ne-jˈma-na
ne-bˈyañi
FINAL VOWEL DELETION
——
ˈne-jma-n
ˈne-byañ
(Other rules)
aɣˈji-i
ˈnijma-n
ˈnepyañ
‘dig (TR)’
‘his hand’
‘bait’
3. Anejom
na bayan
240
Explaining the Aberrant Austronesian Languages
100 Solomon Islands languages and dialects. However, the sheer number of
languages combined with the scarcity of good descriptions meant that these
tables leave many loose ends.
There have been a number of more detailed studies. Painstaking work on
the development of the Southern Vanuatu languages by John Lynch (1978,
2001) has teased apart their complex phonological histories and shown that
many lexical retentions have been disguised by a series of sound changes.
Table 4 contains a few examples from Lynch (2001) that show the steps by
which POc prototypes were transformed into very different forms in three
Southern Vanuatu languages. “Pre-PSV” here refers to “Pre Proto Southern
Vanuatu”, a stage prior to Proto Southern Vanuatu.
Significant advances have been made in understanding sound changes
in the notoriously difficult New Caledonian languages, e.g., by Geraghty
(1989), Haudricourt (1951, 1971), Lynch and Ozanne-Rivierre (2001),
Ozanne-Rivierre (1982, 1992, 1995), Ozanne-Rivierre and Rivierre (1989)
and Rivierre (1991, 1993). Once the sound changes in these New Caledonian
languages have been unravelled it turns out that cognates are more plentiful
than appears on casual inspection.
Alexandre François has recently shown how languages of the Banks
Islands have by a series of natural changes developed systems of from six to
16 vowels (François 2005). In Rotuman the number of contrasting vowels
doubled from five to ten because of metathesis, specifically, anticipatory
copying of a vowel in the next consonant-vowel syllable, caused unlike vowels
to be adjacent and to fuse (Biggs 1965, Blevins and Garret 1998). Blevins
and Garret (1998) note that Kwara‘ae of Malaita (Sohn 1980) and Ririo of
Choiseul (Layock 1982) are in the process of changing their vowel systems
in ways quite similar to Rotuman. In Oceanic cases of such transformations
are not confined to Melanesian languages. For instance, most Micronesian
languages have developed vowel systems a good deal more complex than that
of POc (e.g., Bender 1968, Dyen 1949).
2.3 Grammatical change
There have been many studies of change in particular grammatical
subsystems in Melanesian languages but most have focused mainly on
relatively conservative languages, e.g., Clark (1973), Evans (2003), Geraghty
(1983), Lichtenberk (1985), Lynch (1996), Pawley (1972, 1973), Ross (1988,
2004a, 2004b) and few have examined single languages in depth. Investigations
of morphosyntactic change undergone by more innovative languages are
fewer. Examples can be found in Chapters 5-7 of Lynch’s (2001) study of the
Southern Vanuatu subgroup, Crowley’s (2000) study of Erromangan verbs,
and in several of the papers in Bril and Ozanne-Rivierre (2004).
Andrew Pawley
241
2.4 Other models to explain extreme lexical change and borrowing
While far-reaching changes in phonology and morphology can often be
readily attributed to system-internal factors, that some Southeast Melanesian
languages retain much less of the POc lexicon than others remains something
of a puzzle. The most extreme degree of lexical replacement is found in Dehu
and Negone of the Loyalties and in several languages of the Santa Cruz area,
particularly Teanu of Vanikoro (Francoise 2006) and the three Reefs-Santa
Cruz languages whose status as Austronesian is disputed (see 2.5). For
these languages the percentages of secure reflexes of PMP reconstructions
on the 200 item list is comparable to that of Kaulong (see 1.8). It is unlikely
that such extreme lexical replacement can be accounted for solely in terms
of system-internal factors. Pawley and Ross (1995) refer to a number of
“external” circumstances and processes likely to have favoured rapid change
in particular Oceanic languages:
(i) Bilingualism between Oceanic and Papuan neighbours, sustained by
intermarriage and trade.
(ii) Taboos on using words coinciding with names of chiefs or the dead.
(iii) Social pressures to maintain or accentuate the distinctiveness of one’s
language (or dialect) from one’s neighbours.
(iv) Small size of speech communities as a result of migration, habitat,
political structures, etc.
(v) Changes of physical environment following migration.
(vi) Cultural changes generated internally or by contact.
Name taboo, especially on using the names of dead chiefs, has been widely
reported for Oceania. Chowning (1985, 1986) attributes rapid lexical change
in a number of languages of New Britain (including Sengseng and Kaulong)
to name taboo and borrowing. Keesing and Fifi‘i (1969) have written on
name tabooing in Kwaio of Malaita, but I do not know of any such studies of
societies in Southeast Melanesia. In my view name taboo is unlikely to have
been a major cause of low lexical retention rates in this region.
Among the above-mentioned factors the one that has received most attention
is bilingualism in Northwest Melanesia—not only Oceanic-Papuan bilingualism
but also among speakers of neighbouring Oceanic languages. It is generally
agreed that sustained bilingualism among small language communities, often
associated with a high frequency of intermarriage and sometimes with the
merging of populations drawn from distinct speech communities, can lead
to large scale lexical borrowing and to radical phonological and grammatical
242
Explaining the Aberrant Austronesian Languages
change and to language shift (Bradshaw 1997; Chowning 1986; Holzknecht
1994; Ross, 1996b, 2001; Thurston 1987, 1994; Tryon 1994).
Within Southeast Melanesia (where Oceanic-Papuan bilingualism is not
attested) there have been insightful studies of lexical borrowing that introduces
useful words (Clark 1982, 1994; Lynch 1994) but also of borrowing that yields
no obvious practical gain (Clark 1982). Rivierre (1994) shows how a number of
languages of northern New Caledonia have enriched their phonemic inventories
through lexical borrowing. A collection of case studies of borrowing among
Oceanic languages can be found in Geraghty and Tent (2004).
Extensive lexical borrowing greatly complicates the task of making sense
of sound correspondences. Grace (1973, 1981, 1992, 1996, 1997) describes
his efforts to establish regular correspondences between Canala and Grand
Couli, two neighbouring languages in southern New Caledonia that appear to
be closely related. Each has 26 consonant phonemes. Canala has 18 vowels
and Grand Couli 14. These inventories of about 40 phonemes are associated
with at least 291 different sound correspondences.
But not all neighbours borrow from one another in such a messy way.
Grace (1992) offers a social explanation for the fact that some Oceanic
languages show relatively regular sound correspondences with closely
related neighbouring languages, while others exhibit highly irregular
correspondences. He suggests that whereas the well-behaved languages
have been spoken by predominantly monolingual communities that have
close-knit social networks with sharply focused linguistic norms, the aberrant
languages have been spoken in a very different social milieu—one where
there are diverse networks crisscrossing different languages and where there
are no well-defined linguistic norms.
William Thurston (1987, 1994) has criticised the “gradualist” Neogrammarian model of linguistic change that has dominated Austronesian
linguistics in the post-Ray/Capell era. Based on his fieldwork among
languages communities in northwest New Britain (one Papuan, the others
Oceanic) he proposes a distinction between two kinds of languages, esoteric
and exoteric, in terms of their social functions. An esoteric language is one
spoken chiefly by insiders, people with the same mother tongue. An exoteric
language may have mother-tongue speakers but it is also much used as a lingua
franca by people who have different mother tongues. The hypothesis is that
the two types of language are prone to change in different ways. Whereas
speakers of an esoteric language tend gradually to add structural complexity
that makes the language more efficient as a vehicle of communication
among native speakers (esterogeny), speakers of an exoteric language tend
to simplify its structure because the language must be easily learned by
outsiders (exoterogeny).
Andrew Pawley 243
At different times in its history a language (L) may change from one to
the other function. During a period when L serves as an exoteric language it
may undergo quite abrupt simplification of structure and lexical items may be
borrowed freely from other languages that are spoken in the area where it is a
lingua franca. Over several millennia a language may go through periods of
relative stability punctuated by episodes of rapid change. Furthermore, from
time to time small communities who speak a complex esoteric language will
choose to abandon their mother tongue in favour of an exoteric language that
is more useful to them because it is a regional lingua franca. It can be seen that
Thurston’s model gives a place to pidginisation, in the sense that a new, simpler
variety of a language develops as a result of its use by non-native speakers.
Thurston seeks to explain the pattern of linguistic diversity in northwest
New Britain largely in terms of these socio-linguistic processes but admits
(Thurston 1994:604) that without a magic window on the past it is hard to
know how well the model accounts for the linguistic history of this or other
regions. Malcolm Ross (pers. comm.) finds that the data Thurston presents
for the Lamogai languages are consistent with his claim of esterogeny, but
his data for Lusi do not support the claim of exoterogeny. For a wide-ranging
review of Thurston (1987) see Bradshaw (1995).
It is reasonable to suppose that changes will spread through very small
speech communities faster than large communities. Furthermore, in small
language communities surrounded by larger ones a high proportion of
spouses are likely to come from outside, speaking a different mother
tongue. It follows that languages that have been through several population
bottlenecks as a result of episodes of migration, warfare, disease, etc. can
be expected to have changed faster than others. The Austronesian dispersal
offers a fine laboratory for studies of population bottleneck effects but most
linguistic studies touching on this matter (e.g., Blust 1991, Pawley 1970, Ross
1991) have not treated lexicon. Clark (1994) and Ozanne-Rivierre (1994)
are exceptions—both look at borrowing by small communities speaking
Polynesian Outlier languages from their Melanesian neighbours. In the case
of Fagauvea, lexical borrowing has led to a spectacular increase in the number
of phonemes (from five vowels to nine and from 10 consonants to 27) and
to a change in syllable structure.
2.5 Blust 2005: Were there pre-Austronesian populations in Southeast
Melanesia?
In a recent review article a leading Austronesian historical linguist does
a surprising thing. Robert Blust (2005) seeks to resurrect the notion that
Papuan languages were spoken in Vanuatu and in what he calls “Southern
Melanesia” (the Loyalties and New Caledonia) long before the arrival of
244
Explaining the Aberrant Austronesian Languages
Oceanic languages. He is led to this view by four observations. First, some
of the languages of Vanuatu and Southern Melanesia have two structural
features that are present in many Papuan languages but are said to be rare in
Austronesian languages spoken elsewhere, except for those spoken in regions
where there are Papuan languages.
Currently accepted views of Pacific prehistory leave the distribution of quinary
number systems and serial verbs in Oc [Oceanic] languages unexplained.
Papuan languages that might have served as sources of these structural features
are spoken in New Guinea, the Bismarck Archipelago, and the Solomons chain
(Bougainville, Vella Lavella, and the Russell islands), but nowhere east or
south of Santa Cruz Island. (Blust 2005:554)
Second, there is a piece of evidence from language family distributions
against the view that pre-Austronesians lacked the seafaring technology to
reach and settle any part of Remote Oceania.
… the Papuan languages of Santa Cruz themselves raise questions that have
never been satisfactorily answered: if the settlement of the Pacific Islands
beyond the Solomons chain required navigational skills that were introduced
with the arrival of AN [Austronesian] speakers, how did Papuan speakers reach
Santa Cruz, which required an open sea crossing of more that 200 miles, with
a small and unknown landfall?
… The simplest conclusion… is that Papuan speakers preceded AN speakers in
Santa Cruz by many millennia, long enough for all traces of their relationship
to languages in western Melanesia to be eradicated by accumulated change.
Blust acknowledges that this conclusion sits awkwardly with the
archaeological evidence.
The problem with this conclusion is that the currently accepted archaeological
baseline in Santa Cruz begins with the arrival of the Lapita pottery complex
(hence speakers of AN languages) no earlier than 3,200 BP (Kirch 2000:94).
In short, the archaeology and the linguistics do not add up to a consistent
interpretation of the settlement history of this area. (Blust 2005:554)
The third and fourth arguments are biological and cultural. Blust assumes
on distributional grounds “that POc speakers were southern Mongoloids
who… lacked such cultural characteristics as piercing of the septum or the
use of penis sheaths” (554-55). However, speakers of Melanesian languages
generally have physical traits, such as dark skins and frizzy hair, which
resemble those of Papuan speakers. On Espiritu Santo and Malakula in
Vanuatu, he writes:
Andrew Pawley 245
the prominent noses and full beards of many men are strikingly similar to
features common among New Guinea highlanders. These physical traits agree
with cultural traits that in some cases are highly distinctive in global perspective,
as the insertion of decorative objects through the pierced septum and the use of
penis sheaths in both northern Vanuatu and interior New Guinea. (554-55)
Blust (2005) considers three possible explanations of the presence of
“rather distinctively Papuan traits of phenotype, language, and culture” in
areas where Papuan languages are not spoken.
1. All Oceanic speakers who moved eastward into the uninhabited Pacific
had acquired these traits through contact with Papuan communities in
western Melanesia.
2. The preceding, but both “mixed” and “unmixed” Oceanic speakers moved
out of western Melanesia together.
3. Oceanic speakers in central and southern Melanesia acquired them from
Papuan speakers via contact in situ.
He rejects the first and second hypotheses. Hypothesis 1 cannot be correct
because if these Papuan physical, cultural and linguistic traits were carried
by the first Oceanic speakers to settle Remote Oceania some 3000 years ago
they would have been part of linguistic communities ancestral to those of
Micronesia and Polynesia. But this is not the case, because Papuan phenotypic,
cultural and linguistic traits are essentially absent in Micronesia and Polynesia.
Hypothesis 2 is said to be implausible because it requires us to suppose that
the Oceanic speakers who first settled Remote Oceania
… consisted of disparate groups, some showing strong evidence of physical,
cultural, and linguistic contact with Papuans, but others retaining their southern
Mongoloid phenotype, decimal counting systems, and so on.… To make an
implausible scenario even more implausible, only that part of the migrating
population that was southern Mongoloid and that lacked Papuan cultural and
linguistic traits was able to reach Micronesia and Polynesia. (555)
That leaves hypothesis 3, that “the Papuan features of language, culture,
and physiognomy” that are common to Oceanic speakers in Vanuatu and
Southern Melanesia must have been acquired by contact in situ. The adoption
of this position, however, leads to a crisis of evidence, because there are no
Papuan languages spoken south of Santa Cruz, and the archaeology of central
and Southern Melanesia has to date yielded no indication of a pre-Lapita
population. Yet, Blust concludes:
246
Explaining the Aberrant Austronesian Languages
There is only one obvious way out of this dilemma, and that is to abandon
the prevailing orthodoxy, which holds that the first settlers of all parts of the
Pacific east of the Solomons were the bearers of the Lapita pottery complex,
hence speakers of AN languages. Instead, it appears almost certain that Papuan
languages were spoken in Vanuatu and Southern Melanesia at the time of
initial AN contact, and that the Papuan features in the AN languages of this
region were acquired locally, rather than mysteriously imported from areas
where Papuan languages are still spoken today. The relative sizes of the two
populations may be roughly estimated by the physical and cultural types that
have survived, and the degree of structural convergence in language. In the
Southeast Solomons, where historical populations are relatively light-skinned,
and Papuan linguistic and cultural features are moderate, the early ANspeaking population presumably was relatively large in relation to the Papuan
substrate, while in Espiritu Santo or Malakula in Vanuatu the reverse appears
to be true. Whatever the details, Papuan languages eventually disappeared
throughout Vanuatu and Southern Melanesia, but left traces of their former
presence in the form of typological skewing. (2005:555)
Blust’s reasons for rejecting hypothesis 1 are sound. However, his
arguments for preferring hypothesis 3 to 2 appear to be frail. First, Blust
underestimates the amount of archaeological work done in Vanuatu and
New Caledonia. Both regions have been quite intensively surveyed by
archaeologists over the past 40 years. With every passing year the case for
positing a pre-Lapita occupation of these regions has grown weaker (Bedford
2003, 2006; Sand 2003; Spriggs 1997). If there was such a population, it must
have been small enough to be archaeologically invisible. On these grounds
hypothesis 3 must be rejected as unlikely.
Second, the presumption that the famously aberrant languages of the
Reefs-Santa Cruz area are Papuan is unjustified. Their status as Austronesian
or otherwise has long been in dispute (Lincoln 1978, Wurm 1978) and
remains so, but the availability of better descriptive data now tends to favour
the view that they are fundamentally Oceanic (Ross pers. comm.). As for
positing pre-Austronesian (i.e., Papuan) languages in Vanuatu, the Loyalties
and New Caledonia, Blust is aware of the objection that not a single such
language has survived. He might reasonably point to parallels in parts of
Northwest Melanesia where all or almost all pre-Austronesian languages
have disappeared on certain large islands: none are left in the Admiralties,
only one on New Ireland and only four in the main Solomons group east of
Bougainville. But the firm view of archaeologists that the Lapita people were
the first permanent settlers of Southeast Melanesia makes the mass extinction
hypothesis considerably less plausible for this region.
The arguments from linguistic typology are also not compelling. There
are two problems with the argument concerning serial verb constructions
Andrew Pawley 247
(SVCs). First, SVCs are not particularly rare in Austronesian languages
outside of Oceanic. Certain types of SVCs are present in Taiwan, in Western
Malayo-Polynesian and Central Malayo-Polynesian, and they must be
attributed to POc itself (Ross 2004b). Second, the types of SVC found in
Vanuatu and Southern Melanesia are structurally unlike those found in the
Papuan languages of New Guinea. The history of serial verb constructions
in Oceanic merits further study.
The case for quinary numeral systems is hard to evaluate. Blust observes
that quinary systems are absent from Micronesia and Polynesia and very rare
in Island Southeast Asia. He points to reasons why quinary systems must
have developed independently, i.e., post-POc, in Oceanic languages of the
New Guinea area and in languages of the Vanuatu-Southern Melanesia area.
One problem here is that it is unclear whether the quinary systems found in
some Papuan languages are original or come from Austronesian sources. I
think of typical counting systems of Papuan languages as (a) one, two, many
systems, and (b) body-parts (not just hands) systems, which almost always
have uneven base numbers. Two possibilities merit further consideration: (i)
That quinary systems existed in POc alongside decimal systems, and (ii) that
quinary counting systems spread into parts of Vanuatu and southern Melanesia
some time after Lapita settlement of the region.
We are left to consider hypothesis 2: that the bearers of the Lapita culture
who settled Southeast Melanesia were a diverse population, including
individuals of both “Southern Mongoloid” and “Papuan” pehnotype. The
notion that Lapita canoes from the Bismarck Archipelago carried some people
of “Papuan” as well as “Southern Mongoloid” phenotype does not seem at
all far-fetched. The Lapita colonisation of Remote Oceania was astonishing
in its speed and scale. Dozens of new settlements were founded in different
parts of Remote Oceania within a century or two, and this must have involved
the movement of considerable numbers of people organised by ambitious and
adventurous leaders. It is hard to imagine that the Oceanic-speaking Lapita
migrants could have accomplished this rapid colonisation without recruiting
men and women from non-Oceanic speaking communities that they came in
contact with in Northwest Melanesia. Whether they were recruited as spouses,
as slaves or in some other role, we may never know. That is not to say every
Lapita canoe setting sail for Remote Oceania carried some passengers of
Papuan stock, only that some vessels did.
Why did the Southern Mongoloid phenotype predominate in the founding
populations of Western Polynesia? Perhaps the first Lapita canoes to reach
Remote Oceania contained people who were of this phenotype. (One school
of thought favours the view that the first Lapita movements bypassed the
main Solomons group (Sheppard and Walter 2006), but evidence to the
248
Explaining the Aberrant Austronesian Languages
contrary is emerging (Felgate 2001).) At any rate, this was the population that
reached Tonga, presumably via Vanuatu and Fiji. As movements of Lapita
peoples from Northwest Melanesia to Southeast Melanesia continued over
the next few generations it may be that canoes carrying mixed populations
came to predominate. At some point, though not necessarily in the Lapita
era, populations of mixed descent reached the main islands of Fiji but not
Rotuma or Western Polynesia. Comparative evidence from Lapita skeletal
material, together with other evidence from population genetics, may in due
course settle this issue.10
4. CONCLUSIONS: WHAT HAVE WE LEARNT?
Grace has written that the aberrant languages of Melanesia provide a
“special opportunity for Austronesian linguistics to make a theoretical
contribution to general linguistics” (1997:18). By this he means that these
languages seem to have developed in ways that were not foreseen in standard
models of linguistic change and the challenge is to understand what these
ways were.
What have we learnt? After 150 years of research and debate there are still
many questions to be answered in understanding how the aberrant languages
of Southeast Melanesia got to be the way they are. Still, a great deal has been
learnt on some important issues. It has been established that all the languages
in question are members of the Oceanic branch of Austronesian that have,
for whatever reasons, changed much more than the conservative members
of the family. Reconstructions of the phonology, morphology and lexicon
of POc have allowed us to identify the ways in which particular languages
have been conservative or innovative. The phonological histories of some
(by no means all) of the aberrant languages of Southeast Melanesia are now
reasonably well understood in the sense that the most probable sequence of
structural changes has been worked out. There has been rather less progress
in reconstructing the sequence of changes that have produced grammatically
aberrant languages. In regard to basic vocabulary, it has been possible to
measure rather precisely the rates of change of particular languages since the
breakup of Proto Malayo-Polynesian and to show there has been enormous
variability in retention rates among the Melanesian languages.
More progress has been made in understanding what linguistic changes
occurred than in understanding why they occurred. In particular, why have
some languages of Southeast Melanesia changed much, much more than
others? The pidginisation hypothesis has been discredited. And while
bilingualism between autochthonous Papuan speech communities and
immigrant Oceanic communities has been an important factor in Northwest
Andrew Pawley
249
Melanesia there is little reason to think this occurred in Southeast Melanesia.
Thus, scholars have been led to look for social and demographic mechanisms
that may have hastened or constrained linguistic change in particular cases.
Historical linguists are not well equipped to investigate social and demographic
mechanisms but it is fair to say that their best chance of achieving insights in
this domain is by means of systematic studies of contemporary Oceanic speech
communities, using the methods of ethnography and modern sociolinguistics.
However, there remains considerable doubt that such mechanisms are enough
to wholly explain the extent and nature of the change exhibited by the most
pervasively aberrant languages of Southeast Melanesia, such as those of the
Santa Cruz area, and Dehu and Nengone of the Loyalties.
ACKNOWLEDGEMENTS
Robert Blust, Paul Geraghty, George Grace and Malcolm Ross provided many
helpful comments on a draft of this paper. I am also grateful to Peter Bellwood, Joel
Bradshaw, Diana Carroll, Ann Chowning, Frank Lichtenberk, Piet Lincoln, Lisa
Matisoo-Smith, Miriam Meyerhoff, Claire Moyse-Faurie, Françoise Ozanne-Rivierre,
Peter Sheppard and Peter Trudgill for additional comments and suggestions.
NOTES
1.
2.
3.
4.
Green (1991) argues forcefully that it is not very useful to think of “Melanesia”
as a culture area or as a biogeographic region. He points out that a much more
important biogeographic boundary is that between Near Oceania and Remote
Oceania, and that human populations did not cross this boundary until some
three millennia ago.
Although the first published use of the terms “aberrant” and “exemplary” to
refer to Melanesian languages I have found is in Grace (1981), Grace made the
same conceptual distinction in several earlier papers, such as Grace (1971a). He
points out that there are some aberrant Austronesian languages in other regions
but the majority are in Melanesia.
The data for the Solomon Island languages are from Tryon and Hackman (1983).
In addition to the 140 or so Melanesian languages spoken in Southeast Melanesia,
about 60 are spoken in the main Solomons group, 24 in Bougainville, 40 in
New Britain, 18 in New Ireland and Mussau, 29 in the Admiralties and about
150 in New Guinea or offshore Islands, including a number in Irian Jaya that
do not belong to the Oceanic subgroup. It is pointless to claim exactness for
these figures or for estimates for Southeast Melanesia because in many cases
there are no clear grounds for deciding whether two speech traditions should be
called distinct languages or merely dialects. However it is fair to say that these
totals represent conservative estimates. If highly divergent dialects were to be
distinguished consistently the numbers would be much higher.
250
5.
6.
7.
8.
9.
Explaining the Aberrant Austronesian Languages
We now know that humans reached New Guinea, New Britain and New Ireland
around 40,000 years ago (Specht 2005 and various other contributions in Pawley
et al. 2005, also Kirch 2000, Spriggs 1997). In New Guinea alone, according to
the most recent classification, there are no fewer than 18 unrelated families and
a number of isolates (Ross 2005). At least another six families have surviving
representatives in Northern Island Melanesia. It is very likely that at least some
of the Papuan families continue languages that were spoken in New Guinea and
Northern Island Melanesia in the late Pleistocene.
After this paper was completed Diana Carroll (pers. comm.) pointed out to
me that around the same time as Forster William Marsden (1782) tabulated
lists of cognates and provided analysis that demonstrated the existence of the
Austronesian family. He went beyond Forster in noting certain sound changes
holding between the Austronesian languages in his sample and in including some
languages of Melanesia in Austronesian. Marsden’s linguistic work is reviewed
in detail in Carroll (2005).
Ray’s early and middle years as a scholar coincided with an era when diffusionism
was highly fashionable in British ethnology, with scholars preferring to attribute
local differences in custom and language to the effects of migrants rather than
to in situ change.
The term “Oceanic” gained general acceptance as the name for the subgroup
defined by Dempwolff following its use by Wilhelm Milke in the 1950s (Milke
1958a, 1958b, 1961), although it was probably first applied to a linguistic
grouping several decades earlier by Wilhelm Schmidt (1899b).
It might be objected that the sample of Melanesian languages in Dempwolff’s
(1934-38) three volume work was too small for him to make the sweeping
claim that all Melanesian languages belong to the Oceanic subgroup. It is true
that in these three volumes Dempwolff cites evidence from just two Melanesian
witnesses: Fijian and Sa‘a (Malaita). However, such an objection would be
based on a misperception. Dempwolff’s actual sample of Melanesian witnesses
was at least 30 and was taken from many regions. In his final work he chose for
illustrative purposes just two languages for which data were particularly rich: he
found more than 300 reflexes of Proto Austronesian etyma in each of Fijian and
Sa‘a. As Chrétien (1962) has observed, the full range of Melanesian languages
examined by Dempwolff is evident from a reading of his earlier papers on
Austronesian historical phonology, published between 1920 and 1929. On p.164
of volume 2 of Dempwolff 1933-38, 1934, Dempwolff writes, “I investigated
material from all Melanesian languages following the method set forth in this
volume”, and he goes on to say that the innovations shared by Fijian and Sa‘a
“are common to all Melanesian languages” [my translation from the German:
AP]. Of course, Dempwolff did not investigate every single language and he is at
fault here for using “all” rather than identifying his sample precisely. But on the
same page he singles out Tuna and Pala (Bismarck Archipelago), Lau (Malaita),
Mota (N. Vanuatu), Aneityum (S. Vanuatu), Motu (south coast of New Guinea)
and Gedaged (north coast of New Guinea) as languages with relatively large
numbers of reflexes (at least 100-200) that show regular correspondences and
these are among the languages cited in his earlier works.
Andrew Pawley 251
10. A recent paper by Kayser et al. (2006) indicates that on their way through
Melanesia the founding settlers of Polynesia recruited few women but more men
from Papuan populations. Whereas 93.8 percent of mtDNA in Kayser et al’s large
sample of Polynesians is of Asian origin, only 28.3 percent of Polynesian Ychromosomes is of Asian origin. The abundant skeletal material from the Teouma
site in Vanuatu is consistent with this finding in that it contains, predominantly,
robust males and gracile females (Stuart Bedford pers. comm.).
REFERENCES
Bedford, Stuart, 2003. The timing and nature of Lapitia colonization in Vanuatu: The
haze begins to clear. In C. Sand (ed.), Pacific Archaeology: Assessments and
Prospects. Proceedings of the International Conference for the 50th Anniversary of
the First Lapita Conference (July 1952). Les Cahiers de l’Archéologie en NouvelleCaledonie, Volume 15. Noumea: New Caledonia Museum, pp.147-58.
——2006. Pieces of the Vanuatu Puzzle. Archaeology of the North, South and Centre.
Terra Australia series no. 23, Research School of Pacific & Asian Studies.
Canberra: Pandanus Books, Research School of Pacific & Asian Studies,
Australian National University.
Bellwood, Peter, 1997. The Prehistory of the Indo-Malaysian Archipelago. 2nd edition.
Honolulu: University of Hawai‘i Press.
Bellwood, Peter and E. Dizon, 2005. The Batanes archaeological project and the
“out of Taiwan” hypothesis for Austronesian dispersal. Journal of Austronesian
Studies (Taitung, Taiwan), 1:1-32.
Bellwood, Peter and E. Dizon, in press. Austronesian cultural origins: Out of Taiwan,
via the Batanes Islands, and onwards to Western Polynesia. In A. Sanchez-Mazas, R.
Blench, M.D. Ross, I. Peiros and M. Lin (eds), Past Human Migrations in East Asia:
Matching Archaeology, Linguistics and Genetics. London: Routledge Curzon.
Bender, Byron W., 1968. Marshallese phonology. Oceanic Linguistics 8:16-35.
Bender, Byron W., W. Goodenough, F. Jackson, J. Marck, K. Rehg, Ho-min Sohn,
S. Trussell and J. Wang, 2003. Proto Micronesian reconstructions. Oceanic
Linguistics 42(1):1-110, 42(2):271-358.
Biggs, Bruce, 1965. Direct and indirect inheritance in Rotuman. Lingua, 14:383-415.
[Reprinted in Geraghty and Tent 2004.]
Biggs, Bruce and Ross Clark, n.d. POLLEX. Proto-Polynesian Lexical Files. Computer
files. Department of Anthropology, University of Auckland.
Blevins, Juliette and Andrew Garret, 1998. The origins of consonant-vowel metathesis.
Language, 74(3):508-56.
Blust, Robert, 1977. A rediscovered Austronesian comparative paradigm. Oceanic
Linguistics, 16(1):1-51.
——1978. Eastern Malayo-Polynesian: A subgrouping argument. In S. Wurm and L.
Carrington (eds), Second International Conference on Austronesian Linguistics:
Proceedings. Canberra: Pacific Linguistics, pp.181-234.
——1981a. Some remarks on labiovelar correspondences in Oceanic languages. In
J. Hollyman and A. Pawley (eds), Studies in Pacific Languages and Cultures.
Auckland: Linguistic Society of New Zealand, pp.229-54.
252
Explaining the Aberrant Austronesian Languages
——1981b. Variation in Retention Rate among Austronesian Languages. Paper
presented at 3rd International Conference on Austronesian Linguistics, Bali,
January.
——1991. Sound change and migration distance. In R.A. Blust (ed.), Papers on
Austronesian Languages and Ethnolinguistics in Honor of George Grace.
Canberra: Pacific Linguistics, pp.27-42.
——1995a. Austronesian Comparative Dictionary. Computer files, Department of
Linguistics, University of Hawai‘i.
——1995b. The prehistory of the Austronesian-speaking peoples. Journal of World
Archaeology, 9(4):453-510.
——1999. Why lexicostatistics doesn’t work: The “universal constant” hypothesis
and the Austronesian languages. In C. Renfrew, A. McMahon and L. Trask
(eds), Time Depth in Historical Linguistics. Cambridge: MacDonald Institute
for Archaeological Research, pp. 311-32.
——2005. Review of Lynch, Ross and Crowley 2002. Oceanic Linguistics, 44(2):544-58.
Bradshaw, Joel, 1995. How and why do people change their languages? Oceanic
Linguistics, 34(1):191-202.
——1997. The population kaleidoscope: Another factor in the Melanesian diversity vs
Polynesian homogeneity debate. Journal of the Polynesian Society, 106:222-49.
Bril, Isabelle and Françoise Ozanne-Rivierre (eds), 2004. Complex Predicates in
Oceanic Languages: Studies in the Dynamics of Binding and Boundness. Berlin:
Mouton de Gruyter.
Capell, Arthur, 1933. The structure of the Oceanic languages. Oceania, 3(4):418-34.
——1943. The Linguistic Position of South-Eastern Papua. Sydney: Australasian
Medical Publishing Co.
——1962. Oceanic linguistics today. Current Anthropology, 3:371-428.
——1971. The Austronesian languages of Australian New Guinea. In T.E. Sebeok
(ed.), Current Trends in Linguistics, Volume 8. Oceania. The Hague: Mouton,
pp.240-340.
Carroll, Diana, 2005. William Marsden and his Malayo-Polynesian Legacy.
Unpublished Ph.D. thesis, Australian National University.
Chowning, Ann, 1985. Rapid lexical change and aberrant Melanesian languages:
Sengseng and its neighbours. In A. Pawley and L. Carrington (eds), Austronesian
Linguistics at the 15th Pacific Science Congress. Canberra: Pacific Linguistics,
pp.169-98.
——1986. Refugees, traders and other wanderers: The linguistic effects of population
mixing in Melanesia. In Paul Geraghty, Lois Carrington and S.A. Wurm (eds),
FOCAL II: Papers from the Fourth International Conference on Austronesian
Linguistics. Canberra: Pacific Linguistics, pp.407-34.
Chrétien, Douglas, 1962. Comment on Capell 1962. Current Anthropology, 3(4):396-98.
Clark, Ross, 1973. Transitivity and case in Eastern Oceanic languages. Oceanic
Linguistics, 12:559-606.
——1982. “Necessary” and “unnecessary” borrowing. In A. Halim, L. Carrington
and S.A. Wurm (eds), Papers from the Third International Conference on
Austronesian Linguistics. 3 volumes. Canberra: Pacific Linguistics, pp.137-43.
Andrew Pawley
253
——1994. The Polynesian Outliers as a locus of language contact. In T.E. Dutton
and D.T. Tryon (eds), Language Contact and Change in the Austronesian World.
Berlin: Mouton de Gruyter, pp.109-39.
—— n.d. Proto North and Central Vanuatu Reconstructions. Computer files. Institute
of Linguistics, University of Auckland.
Codrington, Robert H., 1885. The Melanesian Languages. Oxford: Clarendon Press.
Crowley, Terry, 2000. How did Erromangan verbs get so messy? In B. Palmer and
P. Geraghty (eds), SICOL. Proceedings of the 2nd International Conference on
Oceanic Linguistics. Volume 2. Historical and Descriptive Studies. Canberra:
Pacific Linguistics, pp.185-99.
Dempwolff, Otto, 1934-38. Vergleichende Lautlehre des Austronesischen Wörtschaftzes:
1. Induktiver Aufbau einer indonesischen Ursprache. ZES 15 (1934). 2. Deduktiver
Anwendung des Urindonesichen auf austronesische Einzelsprachen, ZES 17
(1937). 3. Austronesiches Wörterverzeichnis, ZES 19 (1938).
Dyen, Isidore, 1949. On the history of the Trukese vowels. Language, 25:420-36.
——1965. A Lexicostatistical Classification of the Austronesian Languages. IUPAL
Memoir 19. Baltimore: Waverly Press.
Evans, Bethwyn, 2003. A Study of Valency-Changing Devices in Proto Oceanic.
Canberra: Pacific Linguistics.
Felgate, Matthew, 2001. A Roviana ceramic sequence and the prehistory of Near
Oceania: Work in progress. In G.R. Clark, A.J. Anderson and T. Vunidilo (eds),
The Archaeology of Lapita Dispersal in Oceania. Terra Australis 17. Canberra:
Pandanus Books, Research School of Pacific and Asian Studies, pp.39-60.
Forster, Johann Reinhold, 1778. Observations Made During a Voyage round the World.
London: Robinson. [New edition edited by Nicholas Thomas, Harriet Guest and
Michael Dettelbach, with a linguistics appendix by Karl H. Rensch. Honolulu:
University of Hawai‘i Press, 1996.]
François, Alexandre. 2005. Unraveling the history of the vowels of seventeen North
Vanuatu languages. Oceanic Linguistics, 44(2):443-504.
——2006. Are Vanikoro languages really Austronesian? Paper read at 2nd Conference
on Austronesian Languages and Linguistics, Oxford, June.
Gabelentz, Hans Conon von der, 1861-73. Die melanesichen Sprachen nach ihrem
grammatischen Bau und ihrer Verwandtschaft unter sich und mit den malaiischpolynesischen Sprachen. Abhandlungen der philologisch-historischen Classe der
Königlich Sächsischen Gesellschaft der Wissenschaften 3, 7.
Geraghty, Paul. 1983. The History of the Fijian Languages. Oceanic Linguistics
Special Publication 19. Honolulu.
——1989. The reconstruction of Southern Oceanic. In R. Harlow and R. Hooper
(eds), VICAL 1: Oceanic Languages. Papers from the Fifth International
Conference on Austronesian Linguistics. Auckland: Linguistic Society of New
Zealand, pp.141-56.
Geraghty, Paul and Jan Tent (eds), 2004. Borrowing: A Pacific Perspective. Canberra:
Pacific Linguistics.
Gordon, Matthew and Ian Maddieson, 1999. The phonetics of Ndumbea. Oceanic
Linguistics, 38(2):66-90.
254
Explaining the Aberrant Austronesian Languages
Grace, George W., 1959. The Position of the Polynesian Languages within the
Austronesian (Malayo-Polynesian) Language Family. International Journal of
American Linguistics Memoir 16. Bloomington.
——1961. Austronesian linguistics and culture history. American Anthropologist,
63:359-68.
——1964. Movement of the Malayo-Polynesians 1500 BC to AD 500: The linguistic
evidence. Current Anthropology, 5:361-68, 403-04.
——1966. Austronesian lexicostatistical classification: A review article. Oceanic
Linguistics, 5(1):13-31.
——1967. Effect of heterogeneity on the lexicostatistical test list: The case of
Rotuman. In G.A. Highland et al. (eds), Polynesian Culture History: Essays in
Honor of Kenneth P. Emory. Bishop Museum Special Publication 56. Honolulu,
pp.289-302.
——1971a. Languages of the New Hebrides and Solomon Islands. In T.E. Sebeok
(ed.), Current Trends in Linguistics, Volume 8. Oceania. The Hague: Mouton,
pp.341-58.
——1971b. Notes on the phonological history of the Austronesian languages of the
Sarmi Coast. Oceanic Linguistics, 10(1):11-37.
——1973. Research on the position of the New Caledonian languages: A progress
report. University of Hawaii Working Papers in Linguistics, 5(7):49-62.
——1981. Direct inheritance and the aberrant Melanesian languages. In J. Hollyman
and A. Pawley (eds), Studies in Pacific Languages and Cultures. Auckland:
Linguistic Society of New Zealand, pp.255-68.
——1992. How do languages change (more on “aberrant” languages)? Oceanic
Linguistics, 31(1):115-30.
——1996. Regularity of change in what? In M. Durie and M. Ross (eds), The
Comparative Method Reviewed. New York: Oxford University Press, pp.157-79.
——1997. On the changing context of Austronesian historical linguistics. In C.
Odé and W. Stokhof (eds), Proceedings of the 7th International Conference on
Austronesian Linguistics. Amsterdam: Editions Rodopi, pp.15-32.
Green, Roger C., 1991. Near and Remote Oceania – disestablishing “Melanesia”
in culture history. In A. Pawley (ed.), Man and a Half: Essays in Pacific
Anthropology and Ethnobiology in Honour of Ralph Bulmer. Auckland:
Polynesian Society, pp.491-502.
——2003. The Lapita horizon and traditions—signature for one set of Oceanic
migrations. In C. Sand (ed.), Pacific Archaeology: Assessments and Prospects.
Proceedings of the International Conference for the 50th Anniversary of the
First Lapita Conference (July 1952). Les Cahiers de l’Archéologie en NouvelleCaledonie, Volume 15. Noumea: New Caledonia Museum, pp.95-120.
Haudricourt, André, 1951. Variations parallèles en Mélanesien. Bulletin de la Société
des Linguistique de Paris, 47:140-53.
——1971. New Caledonia and the Loyalty Islands. In T.E. Sebeok (ed.), Current
Trends in Linguistics, Volume 8. Oceania. The Hague: Mouton, pp.359-96.
Hervas y Panduro, 1784. Catalogo delle lingue. Volume 17 of his Idea dell’ Universo.
21 volumes. In Cesena.
Andrew Pawley
255
Holzknecht, Susanne, 1994. The mechanisms of linguistic change in Labu. In T.E.
Dutton and D.T. Tryon (eds), 1994. Language Contact and Change in the
Austronesian World. Berlin: Mouton de Gruyter, pp.351-75.
Kayser, Manfred, S. Brauer and 13 others, 2006. Melanesian and Asian origins of
Polynesians: mtDNA and Y-chromosome gradients across the Pacific. Molecular
Biology and Evolution. (Advance Access version 21.8.06)
Keesing, Roger and Jonathan Fifi‘i, 1969. Kwaio word tabooing in its cultural context.
Journal of the Polynesian Society, 78(2):154-77.
Kern, Henrik, 1886. De Fidjitaal. Amsterdam.
Kirch, Patrick V., 2000. On the Road of the Winds: An Archaeological History of the Pacific
Islands before European Contact. Berkeley: University of California Press.
Laycock, Donald C., 1982. Metathesis in Austronesian: Ririo and other cases. In A.
Halim, L. Carrington and S.A. Wurm (eds), Papers from the Third International
Conference on Austronesian Linguistic. 3 Volumes. Canberra: Pacific Linguistics,
pp.269-81.
Lichtenberk, Frantisek, 1978. A third palatal reflex in Manam. University of Hawaii
Working Papers in Linguistics, 10(1):183-90.
——1985. Possessive construction in Oceanic languages and Proto-Oceanic. In A.
Pawley and L. Carrington (eds), Austronesian Linguistics at the 15th Pacific
Science Congress. Canberra: Pacific Linguistics, pp.93-140.
Lincoln, Peter, 1978. Reef-Santa Cruz as Austronesian. In S. Wurm and L. Carrington
(eds), Second International Conference on Austronesian Linguistics: Proceedings.
Canberra: Pacific Linguistics, pp.929-67.
Lynch, John, 1978. Proto South Hebridean and Central Pacific. In S. Wurm and L.
Carrington (eds), Second International Conference on Austronesian Linguistics:
Proceedings. Canberra: Pacific Linguistics, pp.717-79.
——1994. Melanesian sailors in a Polynesian sea: Maritime vocabulary in southern
Vanuatu. In A. Pawley and M. Ross (eds), Austronesian Terminologies: Continuity
and Change. Canberra: Pacific Linguistics, pp.289-300.
——1996. Proto Oceanic possessive-marking. In J. Lynch and F. Pat (eds), Oceanic
Studies: Proceedings of the First International Conference on Oceanic
Linguistics. Canberra: Pacific Linguistics, pp.93-100.
——2000. Reconstructing Proto Oceanic stress. Oceanic Linguistics, 39(1):53-82.
——2001. The Linguistic History of Southern Vanuatu. Canberra: Pacific Linguistics.
Lynch, John and Françoise Ozanne-Rivierre, 2001. Some shared developments in
pronouns in languages of Southern Oceania. Oceanic Linguistics, 40(1):33-66.
Lynch, John, Malcolm Ross and Terry Crowley, 2002. The Oceanic Languages.
Richmond, Surrey: Curzon.
Marsden, William, 1782. Remarks on the Sumatran language. Letter to Sir Joseph
Banks dated March 5, 1780. Archaeologia, V(6):154-58.
Milke, Wilhelm, 1958a. Zur inneren Gliederung und geschichtlichen Stellung der
ozeanisch-austronesischen Sprachen. Zeitschrift für Ethnologie, 83:58-62.
——1958b. Ozeanische Verwandtschaftsnamen. Zeitschrift für Ethnologie, 83:226-29.
——1961. Beiträge zur ozeanischen Linguistik. Zeitschrift für Ethnologie, 86:162-82.
——1968. Proto-Oceanic addenda. Oceanic Linguistics, 7(2):147-71.
256
Explaining the Aberrant Austronesian Languages
Ozanne-Rivierre, Françoise, 1976. Le iaai: Language mélanesiénne d’Ouvea
(Nouvelle Calédonie). Langues et Civilisations a Tradition Orale, pub. 20. Paris:
SELAF.
——1982. Langues de Hienghène et proto-oceanic: Phonologie comparée. In A-G.
Haudricourt and F. Ozanne-Rivierre (eds), Dictionnaire thématique des languages
de la region de Hienghène (Nouvelle-Calédonie), Paris: SELAF, pp.9-61.
——1992. The proto Oceanic consonantal system and the languages of New
Caledonia. Oceanic Linguistics, 31(2):191-207.
——1994. Iaai loanwords and phonemic changes in Fagauvea. In T.E. Dutton and
D.T. Tryon (eds), Language Contact and Change in the Austronesian World.
Berlin: Mouton de Gruyter, pp.523-49.
——1995. Structural changes in the languages of northern New Caledonia. Oceanic
Linguistics, 34(1):44-72.
Ozanne-Rivierre, Françoise and J-C. Rivierre, 1989. Nazalization/oralization: Nasal
vowel development and consonant shifts in New Caledonian languages. In R.
Harlow and R. Hooper (eds), VICAL 1: Oceanic Languages. Papers from the
Fifth International Conference on Austronesian Linguistics. Auckland: Linguistic
Society of New Zealand, pp.413-32.
Pawley, Andrew, 1970. Grammatical reconstruction and change in Polynesia and Fiji.
In S.A. Wurm and D.C. Laycock (eds), Pacific Linguistic Studies in Honour of
Arthur Capell. Canberra: Pacific Linguistics, pp.301-67.
——1972. On the internal relationships of Eastern Oceanic languages. In R.C. Green
and M. Kelly (eds), Studies in Oceanic Culture History, Volume 3. Pacific
Anthropological Records No.13. Honolulu: Bishop Museum, pp.1-142.
——1973. Some problems in Proto-Oceanic grammar. Oceanic Linguistics, 12:103-88.
——2002. The Austronesian dispersal: Languages, technologies, peoples. In P.
Bellwood and C. Renfrew (eds), Examining the Farming/Language Dispersals
Hypothesis. Cambridge: McDonald Institute of Archaeological Research,
Cambridge University, pp.251-73.
——2003. Locating Proto Oceanic. In M. Ross et al., The Lexicon of Proto Oceanic.
Volume 2. The Physical Environment. Canberra: Pacific Linguistics, pp.17–34.
Pawley, Andrew, R. Attenborough, J. Golson, R. Hide (eds), 2005. Papuan Pasts:
Cultural, Linguistic and Biological Histories of the Papuan-speaking Peoples.
Canberra: Pacific Linguistics.
Pawley, Andrew and Malcolm Ross, 1995. The prehistory of Oceanic languages:
A current view. In P. Bellwood, J. Fox and D. Tryon (eds), The Austronesians:
Historical and Comparative Perspectives. Canberra: Department of Anthropology,
Research School of Pacific and Asian Studies, Australian National University,
pp.39-74.
Ray, Sidney H., 1892-93. The languages of British New Guinea. Transactions of
the 9th International Congress of Orientalists. London 1892-93, pp.754-79.
(Republished 1894 in J. Anthropological Institute, 24:15-39.)
——1926. A Comparative Study of the Melanesian Island Languages. Cambridge:
University Press.
Reland, Hadrian, 1708. Dissertatio de linguis insularum quarundem orientalium,
partes tres. Trajecti ad Rhenum, pp.55-139.
Andrew Pawley 257
Rivierre, Jean-Claude, 1973. Phonologie comparée des dialects de’l’Extreme-Sud de
la Nouvelle-Caledonie. Paris: SELAF.
——1991. Loss of final consonants in the north of New Caledonia. In R.A. Blust (ed.),
Papers on Austronesian Languages and Ethnolinguistics in Honor of George
Grace. Canberra: Pacific Linguistics, pp.415-32.
——1993. Tonogenesis in New Caledonia. In J. Edmondson and K. Gregerson (eds),
Tonality in Austronesian Languages. Oceanic Linguistics Special Publication 24.
Honolulu: University of Hawaii Press, pp.155-73.
——1994. Contact-induced phonological complexification in New Caledonia. In T.E.
Dutton and D.T. Tryon (eds), Language Contact and Change in the Austronesian
World. Berlin: Mouton de Gruyter, pp.497-522.
Ross, Malcolm, 1988. Proto Oceanic and the Austronesian Languages of Western
Melanesia. Canberra: Pacific Linguistics.
——1989. Proto-Oceanic consonant grade and Milke’s */nj/. In R. Harlow and R.
Hooper (eds), VICAL 1: Oceanic Languages. Papers from the Fifth International
Conference on Austronesian Linguistics. Auckland: Linguistic Society of New
Zealand, pp.433-95.
——1991. How conservative are sedentary languages? In R.A. Blust (ed.), Papers
on Austronesian Languages and Ethnolinguistics in Honor of George Grace.
Canberra: Pacific Linguistics, pp.431-51.
——1996a. On the genetic affiliations of the Oceanic languages of Irian Jaya. Oceanic
Linguistics, 35(2):259-71.
——1996b. Contact-induced changes and the comparative method: Cases from
Papua New Guinea. In M. Durie and M. Ross (eds), The Comparative Method
Reviewed. New York: Oxford University Press, pp.180-216.
——2001. Contact-induced change in Oceanic languages in northwest Melanesia.
In R.M.W. Dixon and A.Y. Aikhenwald (eds), Areal Diffusion and Genetic
Inheritance: Problems in Comparative Linguistics. Oxford: Oxford University
Press, pp.134-66.
——2004a. The morphosyntactic typology of Oceanic languages. Languages and
Linguistics, 5(2):491-541.
——2004b. The grammaticization of directional verbs in Oceanic languages. In I.
Bril and F. Ozanne-Rivierre (eds), Complex Predicates in Oceanic Languages:
Studies in the Dynamics of Binding and Boundness. Berlin: Mouton de Gruyter,
pp.297-329.
——2005. Pronouns as a preliminary diagnostic for grouping Papuan languages. In A.
Pawley, R. Attenborough, J. Golson and R. Hide (eds), Papuan Pasts: Cultural,
Linguistic and Biological Histories of the Papuan-Speaking Peoples. Canberra:
Pacific Linguistics, pp.15-65.
Ross, Malcolm, Andrew Pawley and Meredith Osmond, 1998. The Lexicon of Proto
Oceanic. Vol. 1. Material Culture. Canberra: Pacific Linguistics.
——2003. The Lexicon of Proto Oceanic. Volume 2. The Physical Environment.
Canberra: Pacific Linguistics.
—— in prep. Volume 3 Plants; Volume 4, Animals; Volume 5 People and Society.
258
Explaining the Aberrant Austronesian Languages
Sand, Christophe (ed.), 2003. Pacific Archaeology: Assessments and Prospects.
Proceedings of the International Conference for the 50th Anniversary of the
First Lapita Conference (July 1952). Les Cahiers de l’Archéologie en NouvelleCaledonie, Volume 15. Noumea: New Caledonia Museum.
Schmidt, Wilhelm, 1899a. Über das Verhältniss der melanesischen Sprachen zu den
polynesischen und untereinander. Sitzungsberichte der Kaiserland Akademie der
Wissenschaften zu Wien, philosophisch-historische Classe, 141:1-93.
——1899b. Die sprachlichen Verhältisse Oceaniens (Melanesiens, Polynesiens,
Mikronesiens und Indonesiens) in ihrer Bedeutung für die Ethnologie.
Mitteilungen der Anthropologischen Gesellschaft zu Wien, 29:245-58.
Sheppard, Peter and Richard Walter, 2006. A revised model of Solomon Islands culture
history. Journal of the Polynesian Society, 115(1):7-46.
Shutler, Richard J. and Jeffrey Marck, 1975. On the dispersal of the Austronesian
horticulturalists. Archaeology and Physical Anthropology in Oceania, 10:81-113.
Sohn, Ho-min, 1980. Metathesis in Kwara‘ae. Lingua, 52:305-23.
Specht, Jim, 2005. Revisiting the Bismarcks: Some alternative views. In A. Pawley,
R. Attenborough, J. Golson and R. Hide (eds), Papuan Pasts: Cultural, Linguistic
and Biological Histories of the Papuan-Speaking Peoples. Canberra: Pacific
Linguistics, pp.235-88.
Spriggs, Matthew, 1997. The Island Melanesians. Blackwell: Oxford and Cambridge, MA.
Thurston, William R., 1987. Processes of Change in the Languages of North-western
New Britain. Canberra: Pacific Linguistics.
——1994. Renovation and innovation in the languages of north-western New Britain.
In T.E. Dutton and D.T. Tryon (eds), Language Contact and Change in the
Austronesian World. Berlin: Mouton de Gruyter, pp.573-609.
Tryon, Darrell, 1976. New Hebrides Languages: An Internal Classification. Canberra:
Pacific Linguistics.
——1994. Language contact and contact-induced change in the Eastern Outer Islands,
Solomon Islands. In T.E. Dutton and D.T. Tryon (eds), Language Contact and
Change in the Austronesian World. Berlin: Mouton de Gruyter, pp.611-47.
Tryon, Darrell and Brian Hackmann, 1983. Solomon Islands Languages: An Internal
Classification. Canberra: Pacific Linguistics.
Wallace, Alfred Russell, 1869. The Malay Archipelago. London: MacMillan.
Wolff, John, 1974. Proto-Austronesian *R and *D. Oceanic Linguistics, 13(1):77-122.
Wurm, Stephen, 1978. Reef- Santa Cruz: Austronesian but …! In S.Wurm and L.
Carrington (eds), Second International Conference on Austronesian Linguistics:
Proceedings. Canberra: Pacific Linguistics, pp.969-1010.